Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Phase splitter question
I'm putting together a Mullard 5-20 type amplifier around a couple of vintage Dynaco A420 ultralinear transformers. (I posted the original schematic on ABSE). I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. I'm just learning about phase splitters (and I've got a lot to learn). I am impressed by the simplicity of the Concertina splitter and intimidated by the possible drawbacks of the long tail splitter. There are several other possibilities. My goal, of course, is superior sonics, good stability, and reasonable simplicity. Any suggestions? Thanks Jon |
#2
|
|||
|
|||
"Jon Yaeger" wrote in message ... | | I'm putting together a Mullard 5-20 type amplifier around a couple of | vintage Dynaco A420 ultralinear transformers. (I posted the original | schematic on ABSE). | | I'm going to deviate a bit from the classic circuit, e.g. I want to use a | triode input, probably an E88CC cascode, in lieu of the EF86 pentode, | because of pentode problems such as noise. | | I'm just learning about phase splitters (and I've got a lot to learn). I am | impressed by the simplicity of the Concertina splitter and intimidated by | the possible drawbacks of the long tail splitter. | | There are several other possibilities. My goal, of course, is superior | sonics, good stability, and reasonable simplicity. | | Any suggestions? | | Thanks | | Jon | | | | look at ABSE I prefer LTP always ; -- -- .................................................. ........................ Choky Prodanovic Aleksandar YU "don't use force, "don't use force, use a larger hammer" use a larger tube - Choky and IST" - ZM .................................................. ........................... |
#3
|
|||
|
|||
Jon Yaeger wrote:
I'm putting together a Mullard 5-20 type amplifier around a couple of vintage Dynaco A420 ultralinear transformers. (I posted the original schematic on ABSE). I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. Contrary to what you have heard on this NG, the EF86 as a pentode is not noisy & certainly nowhere near what your preamp will do to the complete system noise figure. Better stick to the phase splitter as used in the original 5-20. The split load (concertina) will be hard pressed to provide enough drive for the output bottles at low distortion. Furthermore, it provides little gain. Better not try to reinvent a successful circuit, at least not until you have considerable experience. You will be busy enough getting stability with your NFB hookup. JLS I'm just learning about phase splitters (and I've got a lot to learn). I am impressed by the simplicity of the Concertina splitter and intimidated by the possible drawbacks of the long tail splitter. There are several other possibilities. My goal, of course, is superior sonics, good stability, and reasonable simplicity. Any suggestions? Thanks Jon |
#5
|
|||
|
|||
"Jon Yaeger" wrote in message ... I'm putting together a Mullard 5-20 type amplifier around a couple of vintage Dynaco A420 ultralinear transformers. (I posted the original schematic on ABSE). I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. I'm just learning about phase splitters (and I've got a lot to learn). I am impressed by the simplicity of the Concertina splitter and intimidated by the possible drawbacks of the long tail splitter. There are several other possibilities. My goal, of course, is superior sonics, good stability, and reasonable simplicity. Any suggestions? Thanks Jon I completed a similar project just a couple of months ago. If you decide on a cascode front end, then your choice of an E88CC is a good one. I wanted to try something else, and came across the 6CG7, which is said to be the noval equivalent of the legendary 6SN7. I used the 6CG7 as a mu-follower, and was pleased with its linearity. For the phase splitter, I used a pentode/triode 6U8A (no Miller effect) in common cathode configuration, from a circuit published in a British mag in the early 60's. The amp has 15dB NFB and sounds much bigger than its 10W, with an impressive dynamic, and low noise floor - with ear against speaker grille (Tannoy Canterburys) there is silence. A pal of mine came to hear it, and made me an offer I could not refuse. Iain |
#6
|
|||
|
|||
Jon Yaeger wrote:
in article , John Stewart at wrote on 3/15/05 7:37 PM: Jon Yaeger wrote: I'm putting together a Mullard 5-20 type amplifier around a couple of vintage Dynaco A420 ultralinear transformers. (I posted the original schematic on ABSE). I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. Contrary to what you have heard on this NG, the EF86 as a pentode is not noisy & certainly nowhere near what your preamp will do to the complete system noise figure. Better stick to the phase splitter as used in the original 5-20. The split load (concertina) will be hard pressed to provide enough drive for the output bottles at low distortion. Furthermore, it provides little gain. Better not try to reinvent a successful circuit, at least not until you have considerable experience. You will be busy enough getting stability with your NFB hookup. JLS Thanks John. I've seen the EF86 used in a Heathkit preamp and it sounded OK to me, too. I didn't mean to imply that the only choices were concertina or long tail were the only two choices. I'm looking for ideas. There are a couple of weaknesses with the original Mullard circuit that I'd like to address: 1. The pentode has an output resistance of 100k, which drives 50 pF of input capacitance from the phase splitter, giving a cut-off of 32 kHz. 2. The output stage has an input capacitance of about 30 pF, and the cut-off of this stage is 100 kHz or lower. 3. The driver stage has only 10 dB of overload capability. 4. I think some 30 dB of feedback is used Your point on the concertina is well taken. However, I wonder if I could follow that with a driver stage using a differential pair of triodes coupled to 6SN7s as cathode followers? OTOH, maybe the easiest thing to do would be to adapt the HK Citation II driver circuit or some other stable, high-performance design. Jon , Hi Jon- I bet you have been reading 'Valve Amplifiers' by Morgan Jones! That is an excellent reference by a guy who really knows his stuff. The information given by Jones will help you cut through some of the chaff you will see here in the NG. As you have pointed, out there are many possibilities in respect to the amplifier driver system. My own personal preference based on relative simplicity, cost & performance these daze is a two stage differential amplifier. That would be a 12AX7 followed by a 12AU7 or alternatively a 6SL7GT followed by a 6SN7GT, depending to a degree on cosmetics. Both of the LTP's to be returned to -ve 150 volts. If you like that would be followed by another twin triode as a pair of CF's direct coupled to the output grids with some safety built in to protect the H-K insulation. You will not normally find that in a common amp from the 50's because of the difficulty in those daze of providing the -ve 150 volts. With SS diodes that is no longer an impediment. You can easily get all the HV, both +ve & -ve from the same winding on your PS transformer. The -ve supply need only be about 15 ma. Back in the 50's my favorite driver was something like a 6CB6 followed by a 12AU7 with both sections paralleled as the split load inverter. The circuit was all DC coupled & used a DC NFB loop from the 12AU7 cathodes to the 6CB6 screen. That stabilized the operating point & gave reasonably good performance from a relatively simple circuit. Noise from the 6CB6 (or 6AU6) front end was not a problem at all. There are many better circuits, but cost & simplicity go out the door, while performance of the amp as a whole does not improve that much. Careful attention to the output stage will get you important gains at that point. That would include DC Balance & Bias Adjustment. Another improvement would be an AC Balance adjustment in the driver stage. You will need some test equipment to set those adjustments. Good Luck with your project, John Stewart |
#7
|
|||
|
|||
"John Stewart" wrote in message ... Jon Yaeger wrote: I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. Contrary to what you have heard on this NG, the EF86 as a pentode is not noisy & certainly nowhere near what your preamp will do to the complete system noise figure. You mentioned Morgan Jones in a later missive. He clearly states the opposite to your view in the paragraph above. With CD as a source, massive gain is not needed, so one could use a triode and perhaps achieve the gain required with less feedback and better stability. Iain |
#8
|
|||
|
|||
Iain M Churches wrote:
"John Stewart" wrote in message ... Jon Yaeger wrote: I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. Contrary to what you have heard on this NG, the EF86 as a pentode is not noisy & certainly nowhere near what your preamp will do to the complete system noise figure. You mentioned Morgan Jones in a later missive. He clearly states the opposite to your view in the paragraph above. True enough, simply showing that everyone has a different 'take' on things. I've certainly never had hum & noise problems using even lesser pentodes such as some of the RF sharp-cutoff types as the front end of the main amplifier. A noise problem at that level usually indicates a faulty tube or some kind of circuit problem which needs to be fixed. But in a preamp, then the EF86 as a pentode or triode could be used to advantage. With CD as a source, massive gain is not needed, so one could use a triode I guess I missed something somewhere since I didn't know only a CD would be used as a program source. I any case, what about the tone controls? That will need another stage ahead of the main amp. and perhaps achieve the gain required with less feedback and better stability. Yes, you could do that. My main point here is 'why reinvent the wheel?', in this case an already proven amp, the Mullard 5-20. That is something for an experienced designer to contemplate. Cheers, John Stewart Iain |
#9
|
|||
|
|||
Jon Yaeger wrote: I'm putting together a Mullard 5-20 type amplifier around a couple of vintage Dynaco A420 ultralinear transformers. (I posted the original schematic on ABSE). I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. Did you find that EF86 was noisy? It really should not have been noisy, since the noise caused by the screen in the pentode is enclosed by the NFB loop, and thus reduced. However, triode operation is better, for puristic reasons. I'm just learning about phase splitters (and I've got a lot to learn). I am impressed by the simplicity of the Concertina splitter and intimidated by the possible drawbacks of the long tail splitter. There are several other possibilities. My goal, of course, is superior sonics, good stability, and reasonable simplicity. Any suggestions? Dynaco ST70 uses a pentode input tube which has its anode gain bootstrapped from the concertina phase inverter, CPI. Thus although there is no gain with CPI, there is plenty in the pentode input tube, and thus when FB is applied, the sensitivity is still OK. So they get away with just one 9 pin socket with a triode pentode for the drive amp, and this low cost solution helped keep the bean counters employed. But I prefer all triode input and LTP stages, since they have less thd and more BW. And more drive capability, and symetrical clipping. I tried em all, and I ended up using what you see at my website, mainly medium U triodes for drivers and inputs. Now I like using EL84 in triode in an LTP to drive KT88/6550 outputs. The sound is most dynamic that way. Patrick Turner. Thanks Jon |
#10
|
|||
|
|||
Jon Yaeger wrote: in article , John Stewart at wrote on 3/15/05 7:37 PM: Jon Yaeger wrote: I'm putting together a Mullard 5-20 type amplifier around a couple of vintage Dynaco A420 ultralinear transformers. (I posted the original schematic on ABSE). I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. Contrary to what you have heard on this NG, the EF86 as a pentode is not noisy & certainly nowhere near what your preamp will do to the complete system noise figure. Better stick to the phase splitter as used in the original 5-20. The split load (concertina) will be hard pressed to provide enough drive for the output bottles at low distortion. Furthermore, it provides little gain. Better not try to reinvent a successful circuit, at least not until you have considerable experience. You will be busy enough getting stability with your NFB hookup. JLS Thanks John. I've seen the EF86 used in a Heathkit preamp and it sounded OK to me, too. I didn't mean to imply that the only choices were concertina or long tail were the only two choices. I'm looking for ideas. There are a couple of weaknesses with the original Mullard circuit that I'd like to address: 1. The pentode has an output resistance of 100k, which drives 50 pF of input capacitance from the phase splitter, giving a cut-off of 32 kHz. So use a phase tweaker network, or use a medium U triode with Ra about 10k. Maybe you still need the phase tweaker. 2. The output stage has an input capacitance of about 30 pF, and the cut-off of this stage is 100 kHz or lower. So use a medium U triode, Ra from each anode is 10k, not as high as 12AX7. 3. The driver stage has only 10 dB of overload capability. That is 3 times the input voltage. Its OK. 4. I think some 30 dB of feedback is used Yes, its a lot, and only possible with great OPTs. Leak used a very similar circuit to the 520, and with very poor OPTs, and they used 26 dB NFB. Stability was a problem. Your point on the concertina is well taken. However, I wonder if I could follow that with a driver stage using a differential pair of triodes coupled to 6SN7s as cathode followers? OTOH, maybe the easiest thing to do would be to adapt the HK Citation II driver circuit or some other stable, high-performance design. Try maybe the circuit at http://www.turneraudio.com.au/htmlwe...0ulabinteg.htm With good OPTs, its difficult to get any better, imho. Patrick Turner. Jon , |
#11
|
|||
|
|||
"John Stewart" wrote in message ... Iain M Churches wrote: "John Stewart" wrote in message ... Jon Yaeger wrote: I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. Contrary to what you have heard on this NG, the EF86 as a pentode is not noisy & certainly nowhere near what your preamp will do to the complete system noise figure. You mentioned Morgan Jones in a later missive. He clearly states the opposite to your view in the paragraph above. True enough, simply showing that everyone has a different 'take' on things. I've certainly never had hum & noise problems using even lesser pentodes such as some of the RF sharp-cutoff types as the front end of the main amplifier. A noise problem at that level usually indicates a faulty tube or some kind of circuit problem which needs to be fixed. But in a preamp, then the EF86 as a pentode or triode could be used to advantage. Yes indeed. The reason that Mullard called the EF86 a "low noise pentode" seems to be that it was "low noise" when compared with other pentodes. With CD as a source, massive gain is not needed, so one could use a triode I guess I missed something somewhere since I didn't know only a CD would be used as a program source. No you didn't miss anything. I made an assumption:-) Perhaps I should have written "If CD is to be the source...." as it most often is these days. The 5-20 has an input sensitivity of 200mV as I recall. That well may be too high for Jon's requirements. With a mu follower triode front end one can get an input sensitivity of 0dBm (0.775V) quite easily and enough NFB to reduce the THD and get the output impedance down to a reasonable figure. any case, what about the tone controls? That will need another stage ahead of the main amp.' Tone controls? I thought that they had been banned under an act of parliament:-) Surely we are talking about a power amp here, or am I missing something? I would have thought a stepped attenuator would have been all that would be required, plus an input selector if more than one input is needed. No pre-amp is better than any pre-amp:-) Yes, you could do that. My main point here is 'why reinvent the wheel?', in this case an already proven amp, the Mullard 5-20. Despite its fame. It is not actually as that good, is it? I can understand why Jon wants to make some changes. This group is the right place to discuss them:-) Cordially, Iain |
#12
|
|||
|
|||
Leak used a very similar circuit to the 520, and with very poor OPTs,
and they used 26 dB NFB. Stability was a problem there's a fix to the Leak OPT, which is to take the covers off and rewire it so all the secondary windings are used. In this form it sounds considerably better, believe me. My 'ultimate' circuit for the Leak Stereo 20 is all triode - PPP 6S4A output valves and a 5BK7 diff pair in the input running off the 5v rectifier heaters, and with one of Morgan Jones CCS under it running from a small trannie which also supplies negative fixed bias to the output tubes. Solid state rectification giving a useful higher B+. I've stuffed four chokes under the bonnet for a really good sounding PSU. You wouldn't recognise this amp as a Leak - it sounds fantastic. No pentodes!!!! No global feedback!!!! At last - the Leak I've dreamed of for 40 years is a reality. It even looks better! Andy |
#13
|
|||
|
|||
Patrick Turner wrote: Try maybe the circuit at http://www.turneraudio.com.au/htmlwe...0ulabinteg.htm With good OPTs, its difficult to get any better, imho. Patrick Turner. Jon , Looks like a Mullard 520 clone to me. Although you have DC Balance for the output tubes you still need some AC Balance adjustment in the driver stage to get optimum performance. And an Output Bias adjusment would be nice to allow for tube aging. I guess that would make it a lot better!! The MJE340 in the return for the LTP makes me nervous. I looked at three different copies of the data sheet & nowhere did a spec for collector capacity appear. To much C at the cathodes of the LTP will render it unbalanced at perhaps barely above the audio band. Anybody know what the collector C of the MJE340 is? That way we can get an idea where the circuit fails. JLS |
#14
|
|||
|
|||
Patrick Turner wrote:
Jon Yaeger wrote: I'm putting together a Mullard 5-20 type amplifier around a couple of vintage Dynaco A420 ultralinear transformers. (I posted the original schematic on ABSE). I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. Did you find that EF86 was noisy? It really should not have been noisy, since the noise caused by the screen in the pentode is enclosed by the NFB loop, and thus reduced. However, triode operation is better, for puristic reasons. I'm just learning about phase splitters (and I've got a lot to learn). I am impressed by the simplicity of the Concertina splitter and intimidated by the possible drawbacks of the long tail splitter. There are several other possibilities. My goal, of course, is superior sonics, good stability, and reasonable simplicity. Any suggestions? Dynaco ST70 uses a pentode input tube which has its anode gain bootstrapped from the concertina phase inverter, CPI. Are you referring to their use of the 6AN8, now as rare as unobtanium? The pentode section is a copy of the 6CB6 while the triode is a 6C4 or 1/2 12AU7. The 6CB6 & the 6AN8 are better known as amplifiers in TV IF strips & other TV & FM applications. NTL, they both work well in audio. AFAIK, that circuit you refer to is not boot strapped. It is usually direct coupled from the pentode plate to the triode grid. It does have lots of gain without bootstrapping. Thus although there is no gain with CPI, there is plenty in the pentode input tube, and thus when FB is applied, the sensitivity is still OK. The gain of the concertina is actually about X1.8 if you take into consideration one input drives two output tubes. JLS So they get away with just one 9 pin socket with a triode pentode for the drive amp, and this low cost solution helped keep the bean counters employed. But I prefer all triode input and LTP stages, since they have less thd and more BW. And more drive capability, and symetrical clipping. I tried em all, and I ended up using what you see at my website, mainly medium U triodes for drivers and inputs. Now I like using EL84 in triode in an LTP to drive KT88/6550 outputs. The sound is most dynamic that way. Patrick Turner. Thanks Jon |
#15
|
|||
|
|||
On Wed, 16 Mar 2005 21:03:59 -0500, John Stewart
wrote: Patrick Turner wrote: Dynaco ST70 uses a pentode input tube which has its anode gain bootstrapped from the concertina phase inverter, CPI. Not the ST70, although other models did. Are you referring to their use of the 6AN8, now as rare as unobtanium? An earlier model, of single channel amplifier. The "Mark II" and "Mark 3" models, with the then-the-rage 6550's and KT88's. Are 6AN8's really so hard to find these days? David probably chose it because it was a common TV tube, and his oft-stated motto was "The cheapest part is the best part". He couldn't, of course, have imagined that motto in today's world of $25 DVD players. Or of $250 old, cooked, ****-poor built, need-all-their-caps-replaced, dangerous to children and pets, no chance in the next three Hells getting UL approval, take your life in your own sorry hands, put yer heads between yer knees and kiss yer sweet ass goodbye, Stereo 70's on Ebay. Modern world. Chris Hornbeck "That's where my forebears came from. Three of them anyway. Who's been sleeping in my porridge?" -F&S |
#16
|
|||
|
|||
John Stewart wrote:
The MJE340 in the return for the LTP makes me nervous. I looked at three different copies of the data sheet & nowhere did a spec for collector capacity appear. To much C at the cathodes of the LTP will render it unbalanced at perhaps barely above the audio band. That is a familiar feeling, isn't it? Most transistor datasheets are quite lacking. Anybody know what the collector C of the MJE340 is? That way we can get an idea where the circuit fails. I believe I have seen af figure of max 25 pF. Regards, Kimjand |
#17
|
|||
|
|||
Andy Evans wrote: Leak used a very similar circuit to the 520, and with very poor OPTs, and they used 26 dB NFB. Stability was a problem there's a fix to the Leak OPT, which is to take the covers off and rewire it so all the secondary windings are used. In this form it sounds considerably better, believe me. The Point one TL12 uses all windings for the various load matches. Stability is still poor no matter how you arrange the secs, although with two parallel 1 ohm windings in series for 4 ohms on the type 2 TL12, you do get the best load match imho, for 8 ohms, and no need to use any other match. But to make the amps unconditionally stable, I had to add in LF gain steps, and for HF, had to add some negative current FB by connecting the botton of the 100 ohms V1 cathode resistor to OV via a 1 uH inductor, which also takes the speaker return curren. Then there is no size of C which would cause oscillation, and a 2 uF C in series with 1.6 ohms, with 16 ohms across the lot to simulate a Quad ESL57 could then be used without causing a huge peak in the sine wave response below 20 kHz. The trouble with Leaks is their high leakage inductance; pardon the pun, but Leak didn't have a clue how to wind OPTs, compared to Williamson, who did. My 'ultimate' circuit for the Leak Stereo 20 is all triode - PPP 6S4A output valves and a 5BK7 diff pair in the input running off the 5v rectifier heaters, and with one of Morgan Jones CCS under it running from a small trannie which also supplies negative fixed bias to the output tubes. Yes but 5 watts of triode from a pair of EL84 is no good unless you have exceptionally sensitive speakers. Solid state rectification giving a useful higher B+. I've stuffed four chokes under the bonnet for a really good sounding PSU. You wouldn't recognise this amp as a Leak - it sounds fantastic. No pentodes!!!! No global feedback!!!! At last - the Leak I've dreamed of for 40 years is a reality. It even looks better! Andy Your'e wecome. There is no need for 4 chokes when 2 max will do if your value for the C is as high at possible, say use standard 470 uF caps and L need only be 4 H. I like the Leak 20 with UL EL84, and with some GFB, but sure, with a decent PS, and with some stability measures. Its still only 12 watts/channel. A pair of 6CM5 in triode could be used to replace the output EL84, and with fixed bias, and 50 mA of idle current, and then you can get 25 watts class AB triode, and that leaves the EL84 for dead. Its doable because the heater current for EL84 is the same as the octal 6CM5. 6FW5 are also excellent, maybe 6AU5. But that means removing the nine pin sockets, and stamping/drilling out the holes to take larger tube sockets, but it will still look nice and tidy when you are done. Patrick Turner. |
#18
|
|||
|
|||
Chris Hornbeck wrote:
On Wed, 16 Mar 2005 21:03:59 -0500, John Stewart wrote: Patrick Turner wrote: Dynaco ST70 uses a pentode input tube which has its anode gain bootstrapped from the concertina phase inverter, CPI. Not the ST70, although other models did. The bootstrapped circuit referred to can be found on p523 of RDH4. I don't recall seeing that in a Dyna, but perhaps. JLS Are you referring to their use of the 6AN8, now as rare as unobtanium? An earlier model, of single channel amplifier. The "Mark II" and "Mark 3" models, with the then-the-rage 6550's and KT88's. Are 6AN8's really so hard to find these days? David probably chose it because it was a common TV tube, and his oft-stated motto was "The cheapest part is the best part". He couldn't, of course, have imagined that motto in today's world of $25 DVD players. Or of $250 old, cooked, ****-poor built, need-all-their-caps-replaced, dangerous to children and pets, no chance in the next three Hells getting UL approval, take your life in your own sorry hands, put yer heads between yer knees and kiss yer sweet ass goodbye, Stereo 70's on Ebay. Modern world. Chris Hornbeck "That's where my forebears came from. Three of them anyway. Who's been sleeping in my porridge?" -F&S |
#19
|
|||
|
|||
John Stewart wrote: Patrick Turner wrote: Try maybe the circuit at http://www.turneraudio.com.au/htmlwe...0ulabinteg.htm With good OPTs, its difficult to get any better, imho. Patrick Turner. Jon , Looks like a Mullard 520 clone to me. The use of SE input tube and LTP make it similar to Mullard 520. But Mullard 520 didn't use low U triodes, nor have a transistor CCS, nor have phase / gain stepping included. Although you have DC Balance for the output tubes you still need some AC Balance adjustment in the driver stage to get optimum performance. There is no need for balancing the AC drive. The CCS ensures the balance of the drive voltages is dead equal if the RL of each LTP are equal. One could of course include AC balance adjustment, but I bet at normal levels the thd might only change marginally, even with very unmatched output tubes. And an Output Bias adjusment would be nice to allow for tube aging. I guess that would make it a lot better!! I normally have an active circuit with a few transistors and leds included in all my amps so that the balance of the output tubes is known at all times, and if the balance is out a bit, one just swings the bias balance pot a bit for the leds to both go out, or have equal brightness, assuming there is an led for each output tube. When one runs out of travel for the adjust pot, its time to renew the tubes. One guy has jan GE6550A which I supplied in 1996, and they are well away from matched now, but still the power is the same as when new, thd the same, and although the grid bias is not the same for the same Ia, the tubes are still serviceable. If a tube thermals out and conducts too much bias current, the active circuit shuts down the B+. My amps are extremely user friendly. The MJE340 in the return for the LTP makes me nervous. I looked at three different copies of the data sheet & nowhere did a spec for collector capacity appear. To much C at the cathodes of the LTP will render it unbalanced at perhaps barely above the audio band. The tiny amount of C at the collector of the MJE340 is in shunt with the very low impedance of the common cathode circuit. The pole caused by the effective RC circuit is well above 100 kHz, and has no effect on the amp operation. Anybody know what the collector C of the MJE340 is? That way we can get an idea where the circuit fails. Its high enough, and with such negligible value, that it has never been a problem in any amp I use this CCS method. By all means measure the Cin at the collector of the MJE340. Also remember that there is an emitter R and that may have some effect on the measured C; I guess you are measuring the collector to base capacitance. There is no miller effect, and the base is taken to a fixed voltage. Patrick Turner. JLS |
#20
|
|||
|
|||
John Stewart wrote: Patrick Turner wrote: Jon Yaeger wrote: I'm putting together a Mullard 5-20 type amplifier around a couple of vintage Dynaco A420 ultralinear transformers. (I posted the original schematic on ABSE). I'm going to deviate a bit from the classic circuit, e.g. I want to use a triode input, probably an E88CC cascode, in lieu of the EF86 pentode, because of pentode problems such as noise. Did you find that EF86 was noisy? It really should not have been noisy, since the noise caused by the screen in the pentode is enclosed by the NFB loop, and thus reduced. However, triode operation is better, for puristic reasons. I'm just learning about phase splitters (and I've got a lot to learn). I am impressed by the simplicity of the Concertina splitter and intimidated by the possible drawbacks of the long tail splitter. There are several other possibilities. My goal, of course, is superior sonics, good stability, and reasonable simplicity. Any suggestions? Dynaco ST70 uses a pentode input tube which has its anode gain bootstrapped from the concertina phase inverter, CPI. Are you referring to their use of the 6AN8, now as rare as unobtanium? Yes. The pentode section is a copy of the 6CB6 while the triode is a 6C4 or 1/2 12AU7. The 6CB6 & the 6AN8 are better known as amplifiers in TV IF strips & other TV & FM applications. NTL, they both work well in audio. Indeed. 6U8A could also be used, pin outs permitting, and load values permitting. AFAIK, that circuit you refer to is not boot strapped. It is usually direct coupled from the pentode plate to the triode grid. It does have lots of gain without bootstrapping. I think you will find the anode load of the pentode is a bootstrapped one, and the reason why the pentode gain is as high as it is, and why there is just *one* gain tube in the input/driver amp, when every other amp, mullard, williamson, McIntosh, etc, all have more than one gain tube. Thus although there is no gain with CPI, there is plenty in the pentode input tube, and thus when FB is applied, the sensitivity is still OK. The gain of the concertina is actually about X1.8 if you take into consideration one input drives two output tubes. Indeed, but basically 1.8 x anything is next to no gain in electronics. JLS Patrick Turner. So they get away with just one 9 pin socket with a triode pentode for the drive amp, and this low cost solution helped keep the bean counters employed. But I prefer all triode input and LTP stages, since they have less thd and more BW. And more drive capability, and symetrical clipping. I tried em all, and I ended up using what you see at my website, mainly medium U triodes for drivers and inputs. Now I like using EL84 in triode in an LTP to drive KT88/6550 outputs. The sound is most dynamic that way. Patrick Turner. Thanks Jon |
#21
|
|||
|
|||
Chris Hornbeck wrote: On Wed, 16 Mar 2005 21:03:59 -0500, John Stewart wrote: Patrick Turner wrote: Dynaco ST70 uses a pentode input tube which has its anode gain bootstrapped from the concertina phase inverter, CPI. Not the ST70, although other models did. One ST70 schematic I have uses EF86 with no bootstrapping for gain, then 1/2 12AU7 for CPI. Another has 7199, triode-pentode, but no bootstrapping same as EF86 and 12AU7, and as I recal one ST70 which I heavily modded, there was a bootstrapped RL for the pentode. Are you referring to their use of the 6AN8, now as rare as unobtanium? An earlier model, of single channel amplifier. The "Mark II" and "Mark 3" models, with the then-the-rage 6550's and KT88's. Are 6AN8's really so hard to find these days? David probably chose it because it was a common TV tube, and his oft-stated motto was "The cheapest part is the best part". He couldn't, of course, have imagined that motto in today's world of $25 DVD players. Or of $250 old, cooked, ****-poor built, need-all-their-caps-replaced, dangerous to children and pets, no chance in the next three Hells getting UL approval, take your life in your own sorry hands, put yer heads between yer knees and kiss yer sweet ass goodbye, Stereo 70's on Ebay. Modern world. ST70 come up OK with a fair amount of work. Patrick Turner. Chris Hornbeck "That's where my forebears came from. Three of them anyway. Who's been sleeping in my porridge?" -F&S |
#22
|
|||
|
|||
Patrick Turner wrote:
John Stewart wrote: Patrick Turner wrote: Try maybe the circuit at http://www.turneraudio.com.au/htmlwe...0ulabinteg.htm With good OPTs, its difficult to get any better, imho. Patrick Turner. Jon , Looks like a Mullard 520 clone to me. The use of SE input tube and LTP make it similar to Mullard 520. But Mullard 520 didn't use low U triodes, nor have a transistor CCS, nor have phase / gain stepping included. Although you have DC Balance for the output tubes you still need some AC Balance adjustment in the driver stage to get optimum performance. There is no need for balancing the AC drive. The CCS ensures the balance of the drive voltages is dead equal if the RL of each LTP are equal. One could of course include AC balance adjustment, but I bet at normal levels the thd might only change marginally, even with very unmatched output tubes. You can easily monitor the unbalance AC current in the output by inserting a small resistor of say 10R in the parallelled cathode return lead of the output tubes. Have a look with a scope. What you will see is the even order (2nd, 4th, Etc) harmonic current. I suspect you will be surprised at what you see. The unbalance can be adjusted out with an AC Balance control in the driver circuit. BTW, this is how the amatuer radio guys build a frequency doubler. When completely implemented it is called a 'push-push' doubler. JLS And an Output Bias adjusment would be nice to allow for tube aging. I guess that would make it a lot better!! I normally have an active circuit with a few transistors and leds included in all my amps so that the balance of the output tubes is known at all times, and if the balance is out a bit, one just swings the bias balance pot a bit for the leds to both go out, or have equal brightness, assuming there is an led for each output tube. When one runs out of travel for the adjust pot, its time to renew the tubes. One guy has jan GE6550A which I supplied in 1996, and they are well away from matched now, but still the power is the same as when new, thd the same, and although the grid bias is not the same for the same Ia, the tubes are still serviceable. If a tube thermals out and conducts too much bias current, the active circuit shuts down the B+. My amps are extremely user friendly. The MJE340 in the return for the LTP makes me nervous. I looked at three different copies of the data sheet & nowhere did a spec for collector capacity appear. To much C at the cathodes of the LTP will render it unbalanced at perhaps barely above the audio band. The tiny amount of C at the collector of the MJE340 is in shunt with the very low impedance of the common cathode circuit. The pole caused by the effective RC circuit is well above 100 kHz, and has no effect on the amp operation. Anybody know what the collector C of the MJE340 is? That way we can get an idea where the circuit fails. Its high enough, and with such negligible value, that it has never been a problem in any amp I use this CCS method. By all means measure the Cin at the collector of the MJE340. Also remember that there is an emitter R and that may have some effect on the measured C; I guess you are measuring the collector to base capacitance. There is no miller effect, and the base is taken to a fixed voltage. Patrick Turner. JLS |
#23
|
|||
|
|||
Kim Johan Andersson wrote: John Stewart wrote: The MJE340 in the return for the LTP makes me nervous. I looked at three different copies of the data sheet & nowhere did a spec for collector capacity appear. To much C at the cathodes of the LTP will render it unbalanced at perhaps barely above the audio band. That is a familiar feeling, isn't it? Most transistor datasheets are quite lacking. Anybody know what the collector C of the MJE340 is? That way we can get an idea where the circuit fails. I believe I have seen af figure of max 25 pF. A pentode tube could be used in place of the MJE340, but would not have the really high real effective collector resistance of the transistor. 25 pF sounds about right, and what effect could that have? If Rin at the cathodes of the tube is 500 ohms, the pole between 25 pf and 500 ohms is at 12 MHz, and hardly likely to cause a problem in my amps. I sometimes use a 1k resistor in series with the collector, so the current can be measured easy, and that would also isolate the C. Patrick Turner. Regards, Kimjand |
#24
|
|||
|
|||
John Stewart wrote: Chris Hornbeck wrote: On Wed, 16 Mar 2005 21:03:59 -0500, John Stewart wrote: Patrick Turner wrote: Dynaco ST70 uses a pentode input tube which has its anode gain bootstrapped from the concertina phase inverter, CPI. Not the ST70, although other models did. The bootstrapped circuit referred to can be found on p523 of RDH4. I don't recall seeing that in a Dyna, but perhaps. Yes, I bet that's where Dynaco must have got it from. It works OK, lotsa gain.... Such a circuit has to make about 25 vrms at the pentode anode, and that is converted to +/- 25v at each phase from the anode and cathode of the triode CPI. It does not matter than the distortion in the pentode is quite high, because they figured it'd never be as high as the distortion in the output stage, and when global NFB is applied back to the cathode of the pentode, all distortions, both the input stages' and the output statges' and the OPT's is all reduced by 10 times if 20 dB of NFB is used. So if the pentode makes 2%, the output tubes make 4%, and the OPT makes 1%, the total is the sq.rt pof the sum of the squared amounts, so total thd = 4.58% with no FB, and with 20 dB of NFB the thd might = 0.5%, and that's a typical outcome. The contribution of the input tube's Dn is thus quite low. Better practice is to use all triodes though, and the spectra is cleaner, and thd at lowest possible. Patrick Turner. JLS Are you referring to their use of the 6AN8, now as rare as unobtanium? An earlier model, of single channel amplifier. The "Mark II" and "Mark 3" models, with the then-the-rage 6550's and KT88's. Are 6AN8's really so hard to find these days? David probably chose it because it was a common TV tube, and his oft-stated motto was "The cheapest part is the best part". He couldn't, of course, have imagined that motto in today's world of $25 DVD players. Or of $250 old, cooked, ****-poor built, need-all-their-caps-replaced, dangerous to children and pets, no chance in the next three Hells getting UL approval, take your life in your own sorry hands, put yer heads between yer knees and kiss yer sweet ass goodbye, Stereo 70's on Ebay. Modern world. Chris Hornbeck "That's where my forebears came from. Three of them anyway. Who's been sleeping in my porridge?" -F&S |
#25
|
|||
|
|||
Patrick Turner wrote:
John Stewart wrote: Chris Hornbeck wrote: On Wed, 16 Mar 2005 21:03:59 -0500, John Stewart wrote: Patrick Turner wrote: Dynaco ST70 uses a pentode input tube which has its anode gain bootstrapped from the concertina phase inverter, CPI. Not the ST70, although other models did. The bootstrapped circuit referred to can be found on p523 of RDH4. I don't recall seeing that in a Dyna, but perhaps. Yes, I bet that's where Dynaco must have got it from. It works OK, lotsa gain.... Such a circuit has to make about 25 vrms at the pentode anode, and that is converted to +/- 25v at each phase from the anode and cathode of the triode CPI. It does not matter than the distortion in the pentode is quite high, because they figured it'd never be as high as the distortion in the output stage, and when global NFB is applied back to the cathode of the pentode, all distortions, both the input stages' and the output statges' and the OPT's is all reduced by 10 times if 20 dB of NFB is used. So if the pentode makes 2%, the output tubes make 4%, and the OPT makes 1%, the total is the sq.rt pof the sum of the squared amounts, so total thd = 4.58% True enough, but all those things hooked together gets in the way & the intermodulation products get much worse. Better to get the THD of each stage down as much as possible before the loop NFB is applied. That bootstrapped pentode driver is a problem looking to happen. JLS with no FB, and with 20 dB of NFB the thd might = 0.5%, and that's a typical outcome. The contribution of the input tube's Dn is thus quite low. Better practice is to use all triodes though, and the spectra is cleaner, and thd at lowest possible. Patrick Turner. JLS Are you referring to their use of the 6AN8, now as rare as unobtanium? An earlier model, of single channel amplifier. The "Mark II" and "Mark 3" models, with the then-the-rage 6550's and KT88's. Are 6AN8's really so hard to find these days? David probably chose it because it was a common TV tube, and his oft-stated motto was "The cheapest part is the best part". He couldn't, of course, have imagined that motto in today's world of $25 DVD players. Or of $250 old, cooked, ****-poor built, need-all-their-caps-replaced, dangerous to children and pets, no chance in the next three Hells getting UL approval, take your life in your own sorry hands, put yer heads between yer knees and kiss yer sweet ass goodbye, Stereo 70's on Ebay. Modern world. Chris Hornbeck "That's where my forebears came from. Three of them anyway. Who's been sleeping in my porridge?" -F&S |
#26
|
|||
|
|||
John Stewart wrote: Patrick Turner wrote: John Stewart wrote: Patrick Turner wrote: Try maybe the circuit at http://www.turneraudio.com.au/htmlwe...0ulabinteg.htm With good OPTs, its difficult to get any better, imho. Patrick Turner. Jon , Looks like a Mullard 520 clone to me. The use of SE input tube and LTP make it similar to Mullard 520. But Mullard 520 didn't use low U triodes, nor have a transistor CCS, nor have phase / gain stepping included. Although you have DC Balance for the output tubes you still need some AC Balance adjustment in the driver stage to get optimum performance. There is no need for balancing the AC drive. The CCS ensures the balance of the drive voltages is dead equal if the RL of each LTP are equal. One could of course include AC balance adjustment, but I bet at normal levels the thd might only change marginally, even with very unmatched output tubes. You can easily monitor the unbalance AC current in the output by inserting a small resistor of say 10R in the parallelled cathode return lead of the output tubes. Have a look with a scope. What you will see is the even order (2nd, 4th, Etc) harmonic current. I suspect you will be surprised at what you see. The unbalance can be adjusted out with an AC Balance control in the driver circuit. BTW, this is how the amatuer radio guys build a frequency doubler. When completely implemented it is called a 'push-push' doubler. JLS Its right what you say, but what the CRO shows at a commoned Rk is the common mode currents in the OPT as a result of the tubes trying to cancel out each other's 2H currents, which never is perfect. However each output tube at most normal listening levels of a watt or two are acting as SE tubes in class A, and their poor matching simply creates an amount of un-cancelled 2H in the thd, and an amount usually far lower than if you used the same two tubes in an SE design with the tubes in parallel. During an upgrade of Quad II amps last year I measured both amps of a pair after replacing all the R&C, and fitting separate RC for cathode, rather than having the common 180 ohms and C from the tertiary winding CT to oV. THis balances the DC pretty well, and far better than Walker allowed us to. The tubes in the amps i fixed were near new, but unmatched, and the owner had a large box full of them. After measuring the amp thd, and finding large amounts of 2H, I swapped the outputs around for least thd, and swapped the EF86 around, and reduced thd 12 dB at 2 watts in one amp and about 15 dB in the other, just exploiting natural non-matches. But where I use all new tubes in my own amps, its rare the 2H ever rises above the 3H, which is low at any level, and almost nothing is to be gained by swapping tubes around. Leak proclaimed his amps would give 0.1% at full power, and that matched tubes were not required, so that if you blow one tube there is no need to buy a pair. He was right. Allen Wright says its better to have a CCS to sink the current from a pair of commoned cathodes of output tubes in PP. This gives maximum cancellation effect but prevents class AB. Only class A is possible. He says the sound quality justifies the technique, but I wonder, since the measurements may predict otherwise. ( but it does allow one output to be grounded, and the other used for a single input phase; its a good way for phase inversion within the output stage, since the output stage becomes a long tail pair, but with magnetically locked RLs. Also a choke can be used for the common RK or current sink. The Williamson uses a common Rk, which also maximises even order current Dn cancelling. I doubt it necessary to have anything other than two grounded cathodes for a fixed bias class AB amp, it seems quite OK imho. Patrick Turner. And an Output Bias adjusment would be nice to allow for tube aging. I guess that would make it a lot better!! I normally have an active circuit with a few transistors and leds included in all my amps so that the balance of the output tubes is known at all times, and if the balance is out a bit, one just swings the bias balance pot a bit for the leds to both go out, or have equal brightness, assuming there is an led for each output tube. When one runs out of travel for the adjust pot, its time to renew the tubes. One guy has jan GE6550A which I supplied in 1996, and they are well away from matched now, but still the power is the same as when new, thd the same, and although the grid bias is not the same for the same Ia, the tubes are still serviceable. If a tube thermals out and conducts too much bias current, the active circuit shuts down the B+. My amps are extremely user friendly. The MJE340 in the return for the LTP makes me nervous. I looked at three different copies of the data sheet & nowhere did a spec for collector capacity appear. To much C at the cathodes of the LTP will render it unbalanced at perhaps barely above the audio band. The tiny amount of C at the collector of the MJE340 is in shunt with the very low impedance of the common cathode circuit. The pole caused by the effective RC circuit is well above 100 kHz, and has no effect on the amp operation. Anybody know what the collector C of the MJE340 is? That way we can get an idea where the circuit fails. Its high enough, and with such negligible value, that it has never been a problem in any amp I use this CCS method. By all means measure the Cin at the collector of the MJE340. Also remember that there is an emitter R and that may have some effect on the measured C; I guess you are measuring the collector to base capacitance. There is no miller effect, and the base is taken to a fixed voltage. Patrick Turner. JLS |
#27
|
|||
|
|||
John Stewart wrote: Patrick Turner wrote: John Stewart wrote: Chris Hornbeck wrote: On Wed, 16 Mar 2005 21:03:59 -0500, John Stewart wrote: Patrick Turner wrote: Dynaco ST70 uses a pentode input tube which has its anode gain bootstrapped from the concertina phase inverter, CPI. Not the ST70, although other models did. The bootstrapped circuit referred to can be found on p523 of RDH4. I don't recall seeing that in a Dyna, but perhaps. Yes, I bet that's where Dynaco must have got it from. It works OK, lotsa gain.... Such a circuit has to make about 25 vrms at the pentode anode, and that is converted to +/- 25v at each phase from the anode and cathode of the triode CPI. It does not matter than the distortion in the pentode is quite high, because they figured it'd never be as high as the distortion in the output stage, and when global NFB is applied back to the cathode of the pentode, all distortions, both the input stages' and the output statges' and the OPT's is all reduced by 10 times if 20 dB of NFB is used. So if the pentode makes 2%, the output tubes make 4%, and the OPT makes 1%, the total is the sq.rt pof the sum of the squared amounts, so total thd = 4.58% True enough, but all those things hooked together gets in the way & the intermodulation products get much worse. Better to get the THD of each stage down as much as possible before the loop NFB is applied. That bootstrapped pentode driver is a problem looking to happen. JLS Well that's what I reckon, as I stated below. The second order products are low in size, but still there, and in very numerable quantity. With regard to RDH4, and pg 523, If the open loop gain of a 1/2 6J5 was 15, with a total of 40k load then for a total of 50v output, ie, +/-25v, then grid input to CPI is 25 + 3.125v, or 28.125v. The current in R1, 0.25M is 3.125 / 250 = 0.0125 mA. The load which thus appears to the anode of the pentode is 28.125 / 0.0125 = 2.250k, or 2.2Mohms, and gain = ( U x RL ) / ( RL + Ra ), and if U = 3,000, and Ra = 1M, then A = ( 3,000 x 2.2 ) / ( 1 + 2.2 ) = 2,062, so only 0.0136 v input between g and k is needed, and 20 dB NFB means g1 input at the pentode would be 0.136 v, and 0.122v is applied to the cathode as NFB. The current in the bootstrapped RL of the pentode is tiny, so there is negligible local current FB, hence Rk of the pentode can be left unbypassed. The trouble with many circuits in old amps which depend on resistors being equal for equal amplitude phases is that resistors drift, or designers choose wrong values for an LTP, where there is merely a common R tail to the LTP, and one of usually low value, say 10k, where the LTP uses a 6SN7. And when changing LTP tubes, the new tube has slightly different gm, so LTP output voltage is rarely the same in old amps. The use of a CCS in an LTP means that even if you have 1/2 a 12AU7 on one side of the LTP and a pentode on the other, the outputs will still be equal, if the RLs are equal, but there will be some 2H, since the distortion of the disimilar tubes cannot cancel. VTL provide an adjusting pot in one side of the LTP as something a service person adjusts. I have had to extract them like a rotten tooth, since pots in DC carrying circuits all to often go noisy. Patrick Turner. with no FB, and with 20 dB of NFB the thd might = 0.5%, and that's a typical outcome. The contribution of the input tube's Dn is thus quite low. Better practice is to use all triodes though, and the spectra is cleaner, and thd at lowest possible. Patrick Turner. JLS Are you referring to their use of the 6AN8, now as rare as unobtanium? An earlier model, of single channel amplifier. The "Mark II" and "Mark 3" models, with the then-the-rage 6550's and KT88's. Are 6AN8's really so hard to find these days? David probably chose it because it was a common TV tube, and his oft-stated motto was "The cheapest part is the best part". He couldn't, of course, have imagined that motto in today's world of $25 DVD players. Or of $250 old, cooked, ****-poor built, need-all-their-caps-replaced, dangerous to children and pets, no chance in the next three Hells getting UL approval, take your life in your own sorry hands, put yer heads between yer knees and kiss yer sweet ass goodbye, Stereo 70's on Ebay. Modern world. Chris Hornbeck "That's where my forebears came from. Three of them anyway. Who's been sleeping in my porridge?" -F&S |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
Phase splitter w/ no name | Vacuum Tubes | |||
What are they Teaching | Audio Opinions | |||
CCS for Mullard phase splitter | Vacuum Tubes | |||
Blindtest question | High End Audio | |||
Repost: Reason 2.0 on a Celeron 2GHz laptop. | Audio Opinions |