| Home |
| Search |
| Today's Posts |
|
|
|
#1
|
|||
|
|||
Passive RIAA VS feedback RIAA preamp
I have never heard a pasive Riaa preamp (like Cary,AES)so don't know
how to compare with the widely used feedback RIAA preamp in vintage gear that I have lots of. What differences are there? I would think that with the feedback type the frequency response would be extended but the dynamic range a little compressed. My speakers are Quad ESL 57 and Rauna Leira(heavy cement speakers and still competitive with the better monitors). It would be nice to eek out all the resolution I can out of the Quads. Theyre stong point is in the midrange so maybe all zero feedback would suit them. Dennis Selwa |
|
#2
|
|||
|
|||
|
Dennis,
What I think it means is that there is a gain stage followed by a tone-shaping loss stage; ergo, there is no feedback around the RIAA network. BTW, you're not related to that fellow from Brooklyn, are you? Best regards, Jon From: (Dennis Selwa) Organization: http://groups.google.com/ Newsgroups: rec.audio.tubes Date: 3 Aug 2003 16:54:53 -0700 Subject: Passive RIAA VS feedback RIAA preamp I have never heard a pasive Riaa preamp (like Cary,AES)so don't know how to compare with the widely used feedback RIAA preamp in vintage gear that I have lots of. What differences are there? I would think that with the feedback type the frequency response would be extended but the dynamic range a little compressed. My speakers are Quad ESL 57 and Rauna Leira(heavy cement speakers and still competitive with the better monitors). It would be nice to eek out all the resolution I can out of the Quads. Theyre stong point is in the midrange so maybe all zero feedback would suit them. Dennis Selwa |
|
#3
|
|||
|
|||
|
BTW, you're not related to that fellow from Brooklyn, are you?
Jon, Dont read all the postings on RAT (I'm reading more every day though) I dont know anyone from Brooklyn except Harvey Rosenberg (RIP). I looked at Mike Turners schematic. That's close to what I want but with 6SL7, and solid state current source. I wish some of these projects would have a photo showing parts layout and also sourcing for critical parts (RIAA etc.) Im considering a hammond aluminum box or just an aluminum plate. Also thinking about Audio Electronic Supply regulated power supply remotely located. |
|
#4
|
|||
|
|||
|
Dennis Selwa wrote: I have never heard a pasive Riaa preamp (like Cary,AES)so don't know how to compare with the widely used feedback RIAA preamp in vintage gear that I have lots of. What differences are there? I would think that with the feedback type the frequency response would be extended but the dynamic range a little compressed. I have always found passive eq filters with RIAA preamps to be quite blameless. The output can easily be 10vrms at all frequencies from 10 Hz to 20 kHz, with a higher voltage ability at the lower F. It is easy to get this 10v at less than 0.1% thd, and no compression. But all you will ever need from your pre is maybe 1 vrms output to a power amp, and thd is often at 0.01%, and this can be from an SET signal amp. Even lower thd, and higher dynamic range is available from a fully balanced tube amp, and there are schematic samples at Allen Wright's http://www.vacuumstate.com/ I have published an RIAA amp for MM at http://www.turneraudio.com.au/htmlwe...tubepreamp.htm My speakers are Quad ESL 57 and Rauna Leira(heavy cement speakers and still competitive with the better monitors). It would be nice to eek out all the resolution I can out of the Quads. Theyre stong point is in the midrange so maybe all zero feedback would suit them. Indeed. Dennis Selwa Happy RIAAing Patrick Turner. |
|
#5
|
|||
|
|||
|
Hi Patrick Surprised to see you actually pointing people to our web site. The schematic to look at, if I may be helpful to our enquirer, is: http://www.vacuumstate.com/images/FVP5A.gif Please note, that while Patrick's is MM this one is MC capable, it has split EQ and thus the termination impedance for the top part of the RIAA needs to be 50K. Also would need some kind of buffering unless built within same chassis. Joe R. PS: Any chance of seeing you coming to Sydney, Pat? I live near the highway BEFORE you get to Liverpool. Plenty of guys come up from Canberra and find it easy to get here... there is no ^*#%% Sydney traffic to deal with! Maybe get some guys from ASoN and we'd have a good time, got some nice Canberra wine! |
|
#7
|
|||
|
|||
|
Hi Pat Its about time you said hello to the dudes down here at RAT. And here I am! The one I have at my site becomes too noisy if the rated input voltage is less than 0.8 mV, if the input termination is 47k. Maybe with less R at the input a 12AX7 could be OK, but few MC carts have high output AND need a low R termination. In fact I like very low R myself. I often use 10-22R with Lo-Outputs MCs, - that will lower noise realative to 47K, but also lower output, so still giving no gain in S/N. So I approve of... I have recently addopted the use of a fet in the RIAA preamp of mine so that like Allen's design, the fet allows MM and MC. When others copy, that is flattery.. :-) But it does work. What fet you use? I prefer to have all the RIAA passive eq done in the one single network, In some cases there isn't any other choice, but when you can split it makes for easier tweaking. I use an inverse RIAA and aim for sq waves at 100Hz, 1KHz and 10KHz. A perfect sq wave indicates virtual flatness a decade above and below that freq. I must say, and I looked at your circuit, not a great admire of SRPP. It is PP and not that great. The distortion created on the bottom tube is fed to the top. Also it strikes me as always hunting to finds its own equalibrium, look at the DC biasing and it looks like TWO cs in series. The fet with a cascoded triode on top, needs less than 10V, and is now a proven method. It sounds clean, precise and not the softness I hear in SRPP. It's more dynamic, involving and still a lot of tube qualities. I don't travel much, too much work, too little money, you know how it is....... I don't like the traffic in Sydney, sure, where you are its not bad, but where I usually stay when I go up is nth shore, and bleedin terrible. Consider you have an open invitation... when/if you do. Just call. Tremain's Audio Encyclopedia, 1965, has a power amp schematic where the tube driver stages were replaced by fets, using the same supply voltages, and the fets seemed to be fairly high voltage rated types, but with numbers I have never seen around. The author went on to describe how IMD/THD was reduced with fets, but methinks it was due to the effectively higher amount of applied FB, due to the extra fet gain. NFB was SO popular in 1965. Almost certainly true! Betcha after IMD/THD they didn't bother listening for differences? I suspect today anyone worth their salts would have. So at least some things *are* better! :-) Joe Rasmussen |
|
#8
|
|||
|
|||
|
wrote: Hi Pat Its about time you said hello to the dudes down here at RAT. And here I am! The one I have at my site becomes too noisy if the rated input voltage is less than 0.8 mV, if the input termination is 47k. Maybe with less R at the input a 12AX7 could be OK, but few MC carts have high output AND need a low R termination. In fact I like very low R myself. I often use 10-22R with Lo-Outputs MCs, - that will lower noise realative to 47K, but also lower output, so still giving no gain in S/N. So I approve of... If you look at the list of attributes in the Ortophon range of MC at http://www.ortofon.com/html/body_mov...renr=133005 1 then you see that the rated outputs are as low as 125 uV, and that typical impedance matches are as low as 10 ohms. The impedances have to be low, to keep the noise low. I have recently addopted the use of a fet in the RIAA preamp of mine so that like Allen's design, the fet allows MM and MC. When others copy, that is flattery.. :-) But it does work. What fet you use? 2SK369. I use it to drive the cathode of a frame grid pentode strapped as a triode. Then I got a passive RIAA, then a 12AT7 bootstrapped follower. Despite NO effort to match the fets, the gain for two channels was within 1 dB. In fact the fets I bought at USD 90c each were very close. The schematic I use differs from Allen's considerably. Once I found the noise a lot lower than a 12AX7, and that the thd of the whole phono stage at 10 vrms output at 1 kHz was lower than 0.1%, I knew I had something worth trying. A listening test went OK. I recently tested a FVP, and it wasn't any better. One of these days, I'll get an MC, and maybe step up a bit more. I prefer to have all the RIAA passive eq done in the one single network, In some cases there isn't any other choice, but when you can split it makes for easier tweaking. I use an inverse RIAA and aim for sq waves at 100Hz, 1KHz and 10KHz. A perfect sq wave indicates virtual flatness a decade above and below that freq. I test and hand trim all the RIAA filters I come across in amps I service with an inverse filter. Most are way out of whack. One won't get a perfect square wave because the roll off below 20 Hz and above 20 kHz prevent it, but it is substantially square at 1 kHz. and the horizontal parts of the square wave should have no wriggles, between 100hz and 20 kHz. I must say, and I looked at your circuit, not a great admire of SRPP. It is PP and not that great. The distortion created on the bottom tube is fed to the top. Nope, the distortion cancels, like in all PP designs. It is VOLTAGE cancelling, a similar but different animal to CURRENT cancelling, which is the basis for all balanced PP power circuits, and balanced signal amps, and LTPs. with the SRPP, the amount of cancelling is dependant on the load, and in fact, with a high value load, like I have in that amp you see, there is a very small amount of cancelling. 3H is quite negligible. In my revised preamp, I have used fixed bias for the top "follower" tube, via a 1 M resistor. Then the R between the bottom gain tube, and the top tube cathode is between 5k and 10k, and the gain tube is RC coupled to the top tube grid, all similar to the 12AX7 input stage at my website. Doing it this way, the gain tube sees a very much higher value loading, which makes the distortion similar to that when using a CCS, which minimises the thd.. The follower also sees a high load value from the bottom tube, and the cap coupled following stage's load, in my case, 50k, and because of local NFB in the follower, the thd is tiny, and it doesn't get back to the gain tube, because its buffered by the follower. Most SET, say a 12AT7, plate loaded, produce maybe 1% of 2H at 10vrms, but used in the way I do, its 20dB lower. Also it strikes me as always hunting to finds its own equalibrium, look at the DC biasing and it looks like TWO cs in series. Nah, That existing 10 tube preamp circuit of mine exhibits NO undulating DC level at the output when large dynamic changes of signal occur. The only time I have seen this happen is where you load a pentode with a CCS, and the 2H distortion creates a slow moving offset voltage with changing levels, since the gain into the CCS is huge, and the Ra of the pentode is high. The fet with a cascoded triode on top, needs less than 10V, and is now a proven method. It sounds clean, precise and not the softness I hear in SRPP. No softness in SRPP, just precision. I have found the old circuit to be quite blameless, and better than anything else I had compared it with. I have used the SRPP for linestage, tone eq, and all sound great. It's more dynamic, involving and still a lot of tube qualities. All I noticed in my fet input revised circuit is that it was more accurate and involving, by a small margin. But I doubt its due solely to the fet cascode circuit, since I have changed the rest of the stages, to make them more linear. And since the fet plus first triode in the phono stage have to work with such tiny voltages, then they alone certainly don't dominate the improvement to the sound. I have also used a revised power supply, with MUCH more capacitance. This was done because with MC, with a low signal, rail movements are undesirable. I have a carefully worked out CLCRCRCRC network to supply the power to the phono input stage. The rail stability is better now than when I used a regulated supply. The cascode circuit with a triode connected frame grid pentode and fet did present some RF problems during development. I did have to keep all the leads very short, and apply a few 0.1uF ceramics to stop oscillations at what I suspect was above 50 Mhz. I fed the DC heater current in the cascode stage via RF chokes, and grounded the heaters with 0.1 uF caps. The internal screen and +10V grid bias supply for the tube must be carefully bypassed to a local star point on the 0V rail. Experimeters in the group building there own versions of cascode should be aware that you will get RF bothers, and it won't be obvious. Don't try to be smart, and run the top tube of the cascode in pentode, its far too unstable. I don't travel much, too much work, too little money, you know how it is....... I don't like the traffic in Sydney, sure, where you are its not bad, but where I usually stay when I go up is nth shore, and bleedin terrible. Consider you have an open invitation... when/if you do. Just call. Tremain's Audio Encyclopedia, 1965, has a power amp schematic where the tube driver stages were replaced by fets, using the same supply voltages, and the fets seemed to be fairly high voltage rated types, but with numbers I have never seen around. The author went on to describe how IMD/THD was reduced with fets, but methinks it was due to the effectively higher amount of applied FB, due to the extra fet gain. NFB was SO popular in 1965. Almost certainly true! Betcha after IMD/THD they didn't bother listening for differences? I suspect today anyone worth their salts would have. So at least some things *are* better! :-) But in the 60s and 70s, quite a few hybrid circuit ideas came out to either make the driver stage work without tubes, or to get rid of the output tubes, and all I can say is that they wasted their time. I have never heard a hybrid power amp which sounded any better than a fully tubed circuit. I think it works to use a fet as an input device in phono amp. But the minute one tries to use the same fet to handle more than a few volts in an SE circuit, the thd is far higher than a triode's, especially if you are use to 6CG7/6SN7, with CCS loading. Maybe one day somone will invent a fet with triode character, ie, low Rd, and U, and low thd with a CCS load, due to the electrostatic feedback from drain to the source, but methinks there is too much solid ****e in the way. Perhaps the 6C45pi russian hi Gm triode might be an excellent tube to use for a cascode circuit as the bottom device, since its Gm is high. But the idle current must be at least 20 mA, so the top cascoded tube needs to be gutsy, and a pair of 6EJ7 in parallel would do. That will be my next test to get a tube amp which is suitable for MC. The cascode has a distinct advantage over a cascade using a fet, loaded by say 2k then cap coupled into a following normal MM amp. The advantage is less noise, and the noise of the following MM grid input, plus that of the fet's 1 k load spoils the snr outcome. Basically, the fet sees a load of the top tubes RL divided by the tube gain, and its action is that of a grounded grid stage. any input noise in the grid circuit of the tube hardly gets amplified, because the fet acts as a high impedance driver into the low impedance of the tube cathode. in effect, the fet driver is a very high value cathode R to the top tube, and one which is of course unbypassed, so the top tube has very low thd, due the the high amount of local current FB. Nearly all the thd is produced by the fet, and its output voltage is low, which is what keeps the thd low of the stage. The remarkable thing about using a fet to drive a tube in the manner described is that it doesn't matter what triode you use, the gain of the circuit between input to the fet gate, and output to the RL at the tube anode, won't vary much, and gain will always be the fet Gm x RL. In this case, the fet's Gm is about 40 mA/V at 6 mA of idle current. The load seen by the top triode, in my case, 6EJ7, is 54, with RL = about 18k, so the load on the fet looking into the tube cathode is 330 ohms. The fet gain is then only 0.04 x 330 = 13.2. Therefore total gain is 712, and I use a source R to reduce that about 8 dB. But 0.04 x 18k = 720, very close to calculating the long way around. The lower the gain of the fet, the lower the input miller effect, so 6DJ8, 12AT7, and 6EJ7 are ideal top triodes to use. even the 6C45pi would be OK, even with the lower current of 6 mA. The fet thd doesn't dramatically increase with a low load value, as its Gm remains fairly constant over such a small change in drain current since the voltage output at the drain is so small in a phono stage. Anti noise freaks might use a quad of fets in parallele to get 6 dB less input noise. In many ways, I found the fet to be a better device than any of the transistors I tried. Patrick Turner. |
| Reply |
| Thread Tools | |
| Display Modes | |
|
|
Similar Threads
|
||||
| Thread | Forum | |||
| recommendations for riaa phono preamp? | Pro Audio | |||
| Echo Mia-MIDI with a Phono PreAmp or TerraTec DMX 6FIRE 24/96 With Software RIAA? | Tech | |||
| Passive preamp : which poweramp for full dynamics and energy? | High End Audio | |||
| Passive Volume Control (Passive Preamp) Info | High End Audio | |||
| Passive RIAA Filtered Preamp Paradox ? | Vacuum Tubes | |||
Hybrid Mode
