Home |
Search |
Today's Posts |
#1
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
Me again folks. I wrote a new page about using a differential input pair of triodes for a PP amp. See my page at
http://www.turneraudio.com.au/basic-tube-5.html Brick bats, buckets of poop, poisoned arrows, drone delivered ordinance all tolerated, but returned to sender with gusto. Patrick Turner. |
#2
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
"patrick-turner" wrote in message ... Me again folks. I wrote a new page about using a differential input pair of triodes for a PP amp. See my page at http://www.turneraudio.com.au/basic-tube-5.html Brick bats, buckets of poop, poisoned arrows, drone delivered ordinance all tolerated, but returned to sender with gusto. Patrick Turner. Alex: I skimmed this page. Noticed an error in the MJE340 current source circuit in the first schematic. (Correct in the others.) Noticed a typo "oft3en" in the first line of the section "DRIVER STAGE DIFFERENTIAL AMP". Was pleased that the term "differential stage" is widely used, and that arcane "long tail pair" is only occasionally -- sort of being phased out. Did not understand the purpose of long and distracting calculations/speculations about mu and gain... Was it all to point out that the actual mu of 6CG7 is 22, not 20 as quoted in the datasheets? There were no explanation why the tubes with mu around 20 are preferred. Obviously, 12AX7 is not a good candidate because it can not work with a decent current, but 12AT7, 6BQ7, ECC85 seem even better than 6CG7 because of higher mu, larger Gm and about the same current. Again, I did not read the whole article attentively. Perhaps there are some other minor errors... |
#3
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
"patrick-turner" wrote in message
... - show quoted text - Me again folks. I wrote a new page about using a differential input pair of triodes for a PP amp. See my page at http://www.turneraudio.com.au/basic-tube-5.html Brick bats, buckets of poop, poisoned arrows, drone delivered ordinance all tolerated, but returned to sender with gusto. Patrick Turner. Alex: I skimmed this page. Noticed an error in the MJE340 current source circuit in the first schematic. (Correct in the others.) Noticed a typo "oft3en" in the first line of the section "DRIVER STAGE DIFFERENTIAL AMP". OK, I will check again. Was pleased that the term "differential stage" is widely used, and that arcane "long tail pair" is only occasionally -- sort of being phased out. Did not understand the purpose of long and distracting calculations/speculations about mu and gain... Was it all to point out that the actual mu of 6CG7 is 22, not 20 as quoted in the datasheets? The purpose of calculations was to not only point out the µ is NOT always the 20you expect to find as data sheets suggest, but to also point out that Ra is more than the usual 8k0 usually given for where Ia = 10mAdc. What varies most in small triodes is Ra and Gm, and their product = µ. So while comtemplating the use of any small triode the µ, Ra and gm used for gain equations must be correct for the Ia condition and Ea condition. The other thing I wanted to point out was that where you have Va of each triode about equal or less than the highest Vin signal, then the anode load load of the tubes becomes quite different, so the Ea change of each triode becomes very different yet the Ia change is the same, where there is a CCS for commoned cathodes. There would be a limit for how much Vin you could have,and I have not measured what it is. But suppose you had 100.1Vac at V1 input grid, and 100.00Vac at V2 input grid, then differential input signal is 0.1Vac so expect say +/-1.7Vac at the anodes, providing the input signals have the same phase and wave shape. The signal that is common to both grids would be 100Vac. So because there is virtually no Ik change anywhere, you only get the amplified diffential signal. But a slight amount of finite resistance or capacitance between cathodes and -ve rail or 0V will produce a common mode output signal at each anode. The CCS shown could be better, with higher finite resistance. And such an input stage can be used for the balanced outputs of a floating transformer secondary of a line level transformer where there may be a high common mode noise signal. There were no explanation why the tubes with mu around 20 are preferred. With medium µ triodes you get a useful amount of gain, reasonably low Ra, and low THD&IMD,and sound is usually excellent. This is a general idea that is valid for all tube amps. Obviously, 12AX7 is not a good candidate because it can not work with a decent current, but 12AT7, 6BQ7, ECC85 seem even better than 6CG7 because of higher mu, larger Gm and about the same current. But diff gain may be excessive in a power amp where you want 2 stages with fairly equal gain and tube types for simplicity and effectiveness and easy re-tubing in 10 years. I much favour the use of 6CG7 in diff mode driving a pair of trioded EL84 as a balanced amp with highish value "tail" resistance, maybe at least 2Ra of one of the triodes. The two stages have overall gain = say 17 x 17 perhaps = 289. If the OP stage needs say 180Vac grid to grid if it uses a large amount of local CFB, then you need 0.62Vac grid to grid at diff input pair. If the GNFB is say 10dB, the VFB will be about 1.24Vac, and Vin becomes 1.86Vac, and this is quite OK, because most of the error correction is done in the OP stage where the beam tetrode Ra is reduced to say less than 1/2 that of the tubes connected as triodes. But a good input pair can be made using 6922/6DJ8, without any change to R&C parts and open loop gain will rise by +3dB. Using less local output stage NFB means more GNFB can be applied and you end up with an amp like the Williamson with 2Vac sensitivity for 16Watts with trioded KT66, and 20dB GNFB. I have used 12AT7/6AQ8 which will run with Ia = 4mA per triode OK. ECC85 and 6SL7 may also be used, 12AY7, and also 12AX7, if the drive voltage to the next stage isn't very high. The following stage Miller C becomes less of a problem causing slewing as voltage drive is reduced. I have often used 2 paralleled triodes for each side of input diff amp, or use small pentode as triodes, 6BX6, 6EJ7, 6AU6, and small power pentodes like 6AR5 etc, etc, etc, etc. And there really isn't any limit to what might be used - but what we end up with can *ALWAYS* be better than some dopey input amp from 1955 that was designed by a shareholder or bean counter. Again, I did not read the whole article attentively. Perhaps there are some other minor errors... I sure ain't perfect, I'm jist a retired amp-worker. A pair of j-fets as a diff pair can be quite good. The ARC Classic 30 with 2 x 6550 output tubes is a perfect example. But one might use a pair of darlington connected bjts, maybe with emitter R on each side, lower RLc, and that can approximate the same goodness of a 6CG7. I'd rather let tubes handle all the signal if there are output tubes in an amp. KR Audio had a SET 30+30 stereo amp with 2 x KRT100 triodes, like 845, but 5V heater and twice gm. But they used an all solid state driver stage. It don't make sense that one but ppl spent $19,000 here to buy one of those amps. Patrick Turner. |
#4
|
|||
|
|||
Quote:
I built simple adaptors for the 9-pin miniatures & did comparative measurements at similar levels with these adaptors in both circuits. Each has a one K, ½ W resistor in series with the grid leads to calm down any tendency to oscillate at RF. The 6SN7 replacement also has 2.2 K, ½ W resistors in series with the cathodes to get its gain down somewhat more like what it replaces. Some of the results with the 6BQ7 family are attached. Cheers to all. Lots of manual labor here on the acres! |
#5
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
Patrick,
Noticed an error in the MJE340 current source circuit in the first schematic. (Correct in the others.) why do you use only one transistor. Using two would also increase the output impedance.... ----------- V+ | | R LOAD | | | C --------B Q1 | E C | Q2 B-------- E | | Re | | ----------- GND Hp |
#6
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
"HpW-Works" wrote in message ... Patrick, Noticed an error in the MJE340 current source circuit in the first schematic. (Correct in the others.) why do you use only one transistor. Using two would also increase the output impedance.... ----------- V+ | | R LOAD | | | C --------B Q1 | E C | Q2 B-------- E | | Re | | ----------- GND Yes, and one most likely does not need -45V rail. Q2 can be a 5-cent general purpose like BC547. Resistor R can be connected to +B rail, not ro the load (common cathodes). This circuit has two disadvantages: 1) It may oscillate at dozens of MHz with certain unlucky combination of Q1 ans Q2; 2) Regulated current is temperature dependent -- drops as the temperature rises. |
#7
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
HpW-Works Aug 9 (20 hours ago)
Other recipients: Patrick, Noticed an error in the MJE340 current source circuit in the first schematic. (Correct in the others.) why do you use only one transistor. Using two would also increase the output impedance.... ----------- V+ | | R LOAD | | | C --------B Q1 | E C | Q2 B-------- E | | Re | | ----------- GND Hp I have ammended the first schematic on that page Basic Tube 5. It now shows the emitter resistor 2k2 of MJE340 being taken to -45Vdc rail. There is no point at all in using 2 transistors in a series arrangement because with just one bjt, the Rout is well over 1Meg ohm. If it was 100Mohms, the operation difference and mathematics involved would not change enough to warrant using the 2 transistor pair. The beauty of having an active solid state CCS means at least it saves having a 1M x 100Watt resistance taken to a -10,000Vdc rail. Its bound to be dangerous. 1.37Vac across 1M = 0.00137mAac, 1.37uA, so no need to worry about Rk being too low. I have used a larger Re with -100V supply rail to increase the finite R value for CCS. But feel free to use two bjts if you insist. In old days the makers of scientific gear used a pentode between common cathodes of LTP and say -150Vdc rail, and they included an Rk of say 3k0 under the pentode k so that effective Ra' = Ra + [ ( µ + 1 ) x Rk ] and if the pentode was 6BX6, gm = 7mA/V at 10mA, and Ra = 250k, then µ = 1,750, and calculated Ra' would be 250k + ( 1,751 x 3,000 ) = 5,503k ohms, or about 5.5M. The screen must be well bypassed to cathode, and sit at about 0V, with high value R feed from say a +400Vdc elsewhere in the gear. But no manufacturer who made large numbers of amps ever put a pentode CCS in any home audio gear because it was never ever thought to improve the sound or lessen the overall THD measurements of the amp. But Now an MJE340 costs 1/100 of the cost of a 6BX6 if we include the tube and socket, wiring, extra chassis space and labour etc. Patrick Turner. |
#8
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
Hi Patrick, But no manufacturer who made large numbers of amps ever put a pentode CCS in any home audio gear because it was never ever thought to improve the sound or lessen the overall THD measurements of the amp. Just for your records, Cary mades a such differential power amp with tube CCS & without zero NFB: http://www.caryaudio.com/products/classic/CAD211FE.html Hp |
#9
|
|||
|
|||
Quote:
To run AB on the 845s surely no tail there. Cheers, John |
#10
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
John Mentioned.... Just for your records, Cary mades a such differential power amp with tube CCS & without zero NFB: http://www.caryaudio.com/products/classic/CAD211FE.html Hp As usual they are keeping the cct a secret! But 6CA7 seems far too much of a tail for the front end difference 6SL7 amp, so perhaps the 300Bs? To run AB on the 845s surely no tail there. Cheers, John It seems Cary Audio are copying the Chinese made Ming-Da monoblocs which use 2 x 845, 2 x 300B drivers, and 2 x 6SN7 input, but with no CCS tail anywhere. I couldn't find a schematic of the Cary amp so I cannot endorse their methods. I always assume all high end makers include gross shortcomings in their designs because every time a sample lands on my bench for singing lessons, de-noising, and to stop them smoking, I have always had to do serious mods to get them to work properly. I totally rebuilt Ming-Da a few years back and owner was very happy. Using 300B as drivers with very low anode current of about 5mA each may seem stupid but it works OK. The owner found the sound of reformed Ming-Da excellent even though there was a total of 4 amp stages - AND 12dB of global NFB. I was able to get a good 85W in class AB, and with 845 biased with only 35mA and with the B+ at 1,200Vdc. Using 2 x EL34 in triode as drivers would be better, and with a CT choke and choke to anode resistances to allow 20mA Ia and high Z RLdc, and hence wide V swing at low THD, as shown at my website. All too hard for Chinese to figure out, and too hard for Cary either. I cannot respect a marketing ploy where the tube line up is given and the use of constant current source ( EL34 ) is touted without saying where or how. BS is BS as usual. CCS work best in an input stage pair where you have an input signal and FB signal and you just want to amplify the difference. The result is always good sounding, low noise, low THD and perfect balance. Such a stage can also be used to work as a floating balanced input amp as any common mode grid signals can't produce much anode signal because of the high Z tail. Other uses for CCS are for anode loads, and cathode loads. I find 2SK1924 works very well as an anode load for EL34, and to voltage proof them a bit you can run two in series - special trick of mine. Patrick Turner. |
#11
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
"Alex Pogossov" wrote in message ... [...] why do you use only one transistor. Using two would also increase the output impedance.... ----------- V+ | | R LOAD | | | C --------B Q1 | E C | Q2 B-------- E | | Re | | ----------- GND Yes, and one most likely does not need -45V rail. Q2 can be a 5-cent general purpose like BC547. Resistor R can be connected to +B rail, not ro the load (common cathodes). Like you say, with R to B+ you'll remove the error dV+/R from above circuit. But i.s.o. R connected to B+ any V+ will do. B+/V+ CC | | R LOAD | | | C +-------B Q1 | E C | Q2 B-------+ E | | Re | | +--------+-- GND This circuit has two disadvantages: 1) It may oscillate at dozens of MHz with certain unlucky combination of Q1 ans Q2; 2) Regulated current is temperature dependent -- drops as the temperature rises. I've always liked the simplicity and junk box available parts (maybe except Q1) of this type of CCS. 1) If oscillation occurs it can usually be suppressed by a small cap between Q2-C and Q1-E. 2) dI/dT is'nt much, only about -0.3%/°C. With suitable (thermal) placement of Q2 and a typical room ambient temp. variation of +/-5°C you'll end up with a current error of -/+1.5%, insignificant for tube electronics I guess. Nevertheless dI/dT can be improved with little change. V+ -+----+ CC | | | R2 R LOAD | | | | | C | +------B Q1(n) | | E | E | +---B Q2(p) | | C | R3 | Re | | | +----+-------+-- GND Unfortunatly this circuit trades very low drift for not so high Ro, also you'll need stabilized V+. V+ -+----+-------+ CC | | | | R2 R R LOAD | | | | | | | C | | +-------B Q1 | | | E | C C | +---B Q3 Q2 B-------+ | E E | R3 | | Re | +---+---+ | | | | | R4 | | | | +--------+------------+-- GND Unfortunatly again, this one trades low drift and high Ro for the KISS principle by yet another transistor. Any thoughts for a CCS with high Ro and low drift, with junk box available parts and KISSes like a dream? Regards, Gio Re. |
#12
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
2) dI/dT is'nt much, only about -0.3%/°C. With suitable (thermal) placement of Q2 and a typical room ambient temp. variation of +/-5°C you'll end up with a current error of -/+1.5%, insignificant for tube electronics I guess. Nevertheless dI/dT can be improved with little change. V+ -+----+ CC | | | R2 R LOAD | | | | | C | +------B Q1(n) | | E | E | +---B Q2(p) | | C | R3 | Re | | | +----+-------+-- GND Unfortunatly this circuit trades very low drift for not so high Ro, also you'll need stabilized V+. Alex: Once we are on this topic, a minor modification -- recycling of Q1 base current -- can improve Ro, particularly its patr associated with Early effect and nonlinearity of the collector junction capacitance. V+-+----+ CC | | | R2 R LOAD | | | | | C | +------B Q1(n) | | E | E | +---B Q2(p) | | C-------+ R3 | | Re +------------+-- GND In this circuit, Q1 base current variations get almost fully (because of high Q2 beta) injected into Q1 emitter, and, because usually differential emitter resistance of Q1 is much smaller than Re, and because of Q1 beta is also high, this injected current passes to Q1 collector and almost completely cancels Q1 collector current variations otherwise caused by beta modulation and non-linear capacitance. |
#13
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
Alex Pogossov Aug 21 (21 hours ago)
Other recipients: 2) dI/dT is'nt much, only about -0.3%/°C. With suitable (thermal) placement of Q2 and a typical room ambient temp. variation of +/-5°C you'll end up with a current error of -/+1.5%, insignificant for tube electronics I guess. Nevertheless dI/dT can be improved with little change. V+ -+----+ CC | | | R2 R LOAD | | | | | C | +------B Q1(n) | | E | E | +---B Q2(p) | | C | R3 | Re | | | +----+-------+-- GND Unfortunatly this circuit trades very low drift for not so high Ro, also you'll need stabilized V+. Alex: Once we are on this topic, a minor modification -- recycling of Q1 base current -- can improve Ro, particularly its patr associated with Early effect and nonlinearity of the collector junction capacitance. V+-+----+ CC | | | R2 R LOAD | | | | | C | +------B Q1(n) | | E | E | +---B Q2(p) | | C-------+ R3 | | Re +------------+-- GND In this circuit, Q1 base current variations get almost fully (because of high Q2 beta) injected into Q1 emitter, and, because usually differential emitter resistance of Q1 is much smaller than Re, and because of Q1 beta is also high, this injected current passes to Q1 collector and almost completely cancels Q1 collector current variations otherwise caused by beta modulation and non-linear capacitance. Suppose you have this bottom schema used for CCS tail of tubed LTP. And suppose you have the B+ rail = 0V and the GND rail at -50V. Then suppose Q2base = -25Vdc, and all bypassed and stabilised. Then Q2 emitter would be at -24.3V and this feeds Q1 base so Q1 emitter is at -25Vdc, same as Q2 base at -25V. So you have very little voltage between Q2 collector and Q1 emitter and hence very little current, but I guess there may be some, and I have not made the circuit so I dunno how well it works or whether or not it oscillates or behaves strangely when trying to be good. I'll keep your inovations in mind, but the use of a single bjt with unbypassed Re and a stable Vb between base and bottom of Re is all anyone needs to get extremely high Rout, and effects of any non linearity in the Rout character are utterly negligible and completely swamped by inherent distortion in the 2 LTP tubes, even when they are good triodes with their RLa = more than 10Ra. Happy soldering, Patrick Turner. |
#14
Posted to rec.audio.tubes
|
|||
|
|||
Differential input stage in PP tube amps
"Alex Pogossov" wrote in message ... 2) dI/dT is'nt much, only about -0.3%/°C. With suitable (thermal) placement of Q2 and a typical room ambient temp. variation of +/-5°C you'll end up with a current error of -/+1.5%, insignificant for tube electronics I guess. Nevertheless dI/dT can be improved with little change. V+ -+----+ CC | | | R2 R LOAD | | | | | C | +------B Q1(n) | | E | E | +---B Q2(p) | | C | R3 | Re | | | +----+-------+-- GND Unfortunatly this circuit trades very low drift for not so high Ro, also you'll need stabilized V+. Alex: Once we are on this topic, a minor modification -- recycling of Q1 base current -- can improve Ro, particularly its patr associated with Early effect and nonlinearity of the collector junction capacitance. V+-+----+ CC | | | R2 R LOAD | | | | | C | +------B Q1(n) | | E | E | +---B Q2(p) | | C-------+ R3 | | Re +------------+-- GND In this circuit, Q1 base current variations get almost fully (because of high Q2 beta) injected into Q1 emitter, and, because usually differential emitter resistance of Q1 is much smaller than Re, and because of Q1 beta is also high, this injected current passes to Q1 collector and almost completely cancels Q1 collector current variations otherwise caused by beta modulation and non-linear capacitance. Brilliant! Maybe except for 1 remaining flaw. Although Q1/Q2 Vbe drift as a function of ambient temperature is compensated, Vbe drift of Q1 as a function of temperature (rise) caused by device dissipation is not. As long as the load is more or less constant, like in application for a LTP, this is no problem. For application with a varying load close thermal coupling between Q1 and Q2 is necessary as well as a low Rth.j-c for Q1. Regards, Gio Re |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
Tube - Tube - SS - 3 stage hybrid amps question | Vacuum Tubes | |||
INA103 differential stage | Pro Audio | |||
Advice for grounded grid input stage | Vacuum Tubes | |||
Questions Of Borbely Differential Tube Balanced Linestage | Vacuum Tubes | |||
FS: Golden Tube SEP-1 tube line stage | Marketplace |