Triode operation frequently sounds better. However, your original requirement stated somewhere between 30 and 50 W. You can't squeeze 30 W. out of either a pair EL34s or a pair of 7591s that are triode wired. A PP pair of triode wired EI KT90s would yield 30 GOOD sounding W. Unfortunately, properly made KT90s are hard to come by.
Since the NATO bombing, the EI plant has produced GUANO.Pentode mode is the way to maximize power O/P. If you regulate screen grid B+ and use O/P trafos with tertiary CFB windings, pentode mode sounds GOOD and also has a satisfactory damping factor. Please keep in mind that the ultralinear topology is also a form of local NFB. To maximize ultralinear performance, a separate screen grid winding and regulated screen B+ should be employed. Observe please, multi-grid power O/P tubes perform best when g2 B+ is regulated.
I was checking out this web site. The Cayin TA-30 is an interesting integrated amp.
http://www.angelfire.com/biz/bizzyb/CAYINPICS.html
http://www.angelfire.com/biz/bizzyb/CAYIINPICS/TA-30schematic.jpg
It appears to be an interesting design. Doesn't look too complicated. I don't know if anyone has the whole schematic diagram. The power suppply capacitors are huge. Probably blow the fuse when turned on.
I know this has had good reviews. Very similar to the Prima Luna Prologue One. Made by the same people. I guess the Prima Luna has an Auto-Bias circuit.
http://www.angelfire.com/biz/bizzyb/CAYINPICS.html
http://www.angelfire.com/biz/bizzyb/CAYIINPICS/TA-30schematic.jpg
It appears to be an interesting design. Doesn't look too complicated. I don't know if anyone has the whole schematic diagram. The power suppply capacitors are huge. Probably blow the fuse when turned on.
I know this has had good reviews. Very similar to the Prima Luna Prologue One. Made by the same people. I guess the Prima Luna has an Auto-Bias circuit.
The hand drawn Cayin schematic shows paralleled sections of a 12AX7 as the voltage amplifier feeding a long tailed pair (LTP) phase splitter (fundamentally differential). The splitter feeds PP EL34s configured in ultralinear mode.
What the schematic does NOT show is the loop NFB circuitry, including crucial phase compensation. Phase compensation is tied directly to the characteristics of the O/P trafo.
The problem with the B+ PSU isn't popping fuses, as a slow blow type takes care of that. That PSU is instant on and the rail voltage is high enough to make cathode stripping a genuine concern. EL34 service life will be short.
When the 2 sections of a twin triode are paralleled, it's important the sections match closely. Otherwise, current hogging occurs and the sound is not too good.
Notice the asymmetrical load resistors in the splitter. That's necessary because the non-inverting triode is grounded in so far as AC is concerned. Grounded grid triodes yield less gain than common cathode triodes yield. A constant current sink (CCS) in the splitter's tail would force symmetrical behavior and allow both load resistors to be the same value. A CCS also improves power supply rejection (PSR).
What the schematic does NOT show is the loop NFB circuitry, including crucial phase compensation. Phase compensation is tied directly to the characteristics of the O/P trafo.
The problem with the B+ PSU isn't popping fuses, as a slow blow type takes care of that. That PSU is instant on and the rail voltage is high enough to make cathode stripping a genuine concern. EL34 service life will be short.
When the 2 sections of a twin triode are paralleled, it's important the sections match closely. Otherwise, current hogging occurs and the sound is not too good.
Notice the asymmetrical load resistors in the splitter. That's necessary because the non-inverting triode is grounded in so far as AC is concerned. Grounded grid triodes yield less gain than common cathode triodes yield. A constant current sink (CCS) in the splitter's tail would force symmetrical behavior and allow both load resistors to be the same value. A CCS also improves power supply rejection (PSR).
"The hand drawn Cayin schematic shows paralleled sections of a 12AX7 as the voltage amplifier feeding a long tailed pair (LTP) phase splitter (fundamentally differential). The splitter feeds PP EL34s configured in ultralinear mode.
"What the schematic does NOT show is the loop NFB circuitry, including crucial phase compensation. Phase compensation is tied directly to the characteristics of the O/P trafo.
The problem with the B+ PSU isn't popping fuses, as a slow blow type takes care of that. That PSU is instant on and the rail voltage is high enough to make cathode stripping a genuine concern. EL34 service life will be short.
When the 2 sections of a twin triode are paralleled, it's important the sections match closely. Otherwise, current hogging occurs and the sound is not too good.
Notice the asymmetrical load resistors in the splitter. That's necessary because the non-inverting triode is grounded in so far as AC is concerned. Grounded grid triodes yield less gain than common cathode triodes yield. A constant current sink (CCS) in the splitter's tail would force symmetrical behavior and allow both load resistors to be the same value. A CCS also improves power supply rejection (PSR)."
Does that mean Cayin well designed? I would think new designs are better than the ST-70.. If nothing else, being forty years further up the learning curve?
"What the schematic does NOT show is the loop NFB circuitry, including crucial phase compensation. Phase compensation is tied directly to the characteristics of the O/P trafo.
The problem with the B+ PSU isn't popping fuses, as a slow blow type takes care of that. That PSU is instant on and the rail voltage is high enough to make cathode stripping a genuine concern. EL34 service life will be short.
When the 2 sections of a twin triode are paralleled, it's important the sections match closely. Otherwise, current hogging occurs and the sound is not too good.
Notice the asymmetrical load resistors in the splitter. That's necessary because the non-inverting triode is grounded in so far as AC is concerned. Grounded grid triodes yield less gain than common cathode triodes yield. A constant current sink (CCS) in the splitter's tail would force symmetrical behavior and allow both load resistors to be the same value. A CCS also improves power supply rejection (PSR)."
Does that mean Cayin well designed? I would think new designs are better than the ST-70.. If nothing else, being forty years further up the learning curve?
I would say there is little special in that design. It's a commercial product whose 1st objective is to turn a profit. I listed some design shortcomings. Profits would be reduced if they were addressed by the manufacturer. FWIW, I dislike the B+ PSU intensely. SS rectification is quite acceptable, but instant on is NOT. That design, without modification, is a tube eater. Ugh!
When it comes to tube circuitry, there's very little new under the sun. We have access to some better passive parts, but our know how is not better than that of the greats who, sadly, are no longer with us. Fisher, Recklinghausen, Hegeman, Marantz, etc. have MUCH to teach us.
The Prima Luna Prologue One which is basically the same amplifier uses a soft-start circuit. It also has some active auto bias circuit which seems a little strange.
Maybe for a first project I will mess with the ST-70 circuit since so many have built it.
Eli, do you have an opinion about which ST-70 driver board is better than the others?
By the way, you forgot to mention David Hafler, Frank McIntosh.
In the early 90's I went to Walton NY to see the MC275 reissue and met Gordon Gow's son. Nice guy. That's as close as I got to meeting one of the greats.
Maybe for a first project I will mess with the ST-70 circuit since so many have built it.
Eli, do you have an opinion about which ST-70 driver board is better than the others?
By the way, you forgot to mention David Hafler, Frank McIntosh.
In the early 90's I went to Walton NY to see the MC275 reissue and met Gordon Gow's son. Nice guy. That's as close as I got to meeting one of the greats.
labjr said:
labjr, you might find it interesting to compare the Stereophile measurements of the Prima Luna with the measurements of an ST-70 using the DiyTube driver board. (You can find the latter by downloading the manual for that driver board and flipping towards the end.)
Despite the marketing hype (about how the Prima Luna autobias circuit decreases distortion by 50%, etc.), the ST-70/DiyTube with Dynaco iron measures a lot better. It has roughly 30-40% less distortion at 1kHz at similar power levels. Towards the low end, the Prima Luna's distortion levels increase significantly even at relatively low power because its output transformers are designed to a certain price point. A person wouldn't want to run the Prima Luna full range without a subwoofer (well, the increased harmonic distortion at the low end could contribute to a warmer, fatter sound that some may like -- though the Stereophile review points out that odd order harmonics increase as the transformer saturates, which indicates that the sound may get harsher rather than warmer).
The square wave response plots for both amps are effectively identical, except that the DiyTube plot is still clean at ten times the power as the Stereophile square wave plot (10W vs. 1W). (BTW, compare the square wave plots of both amps with the poor square wave plot in the Dynaco STII review from Stereophile -- a modern driver board can clearly clean up the amp's performance.)
Measurements are not everything, of course, but at least these DIY ST-70s are competitive. It's really the quality of the iron that will probably make the most significant difference though. (The low quality of the Prima Luna's iron is not surprising... if you break the Prima Luna's price down (dealer markup, importer markup, import tariffs, manufacturer profit, assembly, parts), you're left with perhaps $50 to buy all three transformers; tough even in China.) The Dyna iron, new or old stock, is very good. For a chuckle, compare the Stereophile graph titled "Prima Luna Prologue One, 8 ohm tap, spectrum of 50Hz sinewave, DC–1kHz, at 3W" versus the Stereophile graph titled "Dynaco Stereo 70 II, spectrum of 50Hz sinewave, DC-1kHz, at 22.5W". At nearly eight times the output power, the ST70 transformers still produce much less distortion, and the distortion spectrum is a lot cleaner too. The 13th order harmonic produced by the Prima Luna will likely be audible and contribute to harshness.
As Wodgy pointed out ST70 O/P "iron" is decent.
All of the replacement ST70 driver boards have their +s and -s. The Triode Electronics board uses the original topology, but bypasses the 7199 issue by using 3 "bottles" (2 pentodes and a twin triode).
The original query asked for an "integrated" amp. All ST70 driver circuits (original and replacement) assume a line stage with gain upstream. Drive voltage requirements for the EL34 are greater than those of the 7591. That's why I suggested the 7591 to begin with. As Fisher, Scott, and EICO (IIRC) demonstrated, the 7591 is a GOOD sounding tube, with power O/P approx. = that of the EL34.
OK, let's assume ultralinear EL34s and Dyna O/P "iron". If a 12AT7 differential splitter ala "El Cheapo" is used, not much additional gain will be needed. Keep the short (splitter and "finals") NFB loop, as the chances of trouble are small. What's left is obtaining a small amount of LOW distortion gain (no loop NFB) at the amp's I/P. There are several ways to tackle that requirement. An interesting possibility is Steve Eddy's work with step up microphone trafos. A voltage follower buffers the level control and provides the low impedance drive trafos "like". The trafo raises the I/P voltage and feeds the inverting grid of the splitter.
All of the replacement ST70 driver boards have their +s and -s. The Triode Electronics board uses the original topology, but bypasses the 7199 issue by using 3 "bottles" (2 pentodes and a twin triode).
The original query asked for an "integrated" amp. All ST70 driver circuits (original and replacement) assume a line stage with gain upstream. Drive voltage requirements for the EL34 are greater than those of the 7591. That's why I suggested the 7591 to begin with. As Fisher, Scott, and EICO (IIRC) demonstrated, the 7591 is a GOOD sounding tube, with power O/P approx. = that of the EL34.
OK, let's assume ultralinear EL34s and Dyna O/P "iron". If a 12AT7 differential splitter ala "El Cheapo" is used, not much additional gain will be needed. Keep the short (splitter and "finals") NFB loop, as the chances of trouble are small. What's left is obtaining a small amount of LOW distortion gain (no loop NFB) at the amp's I/P. There are several ways to tackle that requirement. An interesting possibility is Steve Eddy's work with step up microphone trafos. A voltage follower buffers the level control and provides the low impedance drive trafos "like". The trafo raises the I/P voltage and feeds the inverting grid of the splitter.
Wodgy has provided a nice solution. 1.06 V. sensitivity for full power is OK. The DIYTube board looks good to me. That CCS in the splitter's tail is important. Just put 10 KOhm log. pots. at the amp's I/Ps and connect the wipers to the driver board's I/Ps.
Use a separate bias pot. and cathode test resistor for each tube. The tubes available tend not to be as tightly matched as 1 bias pot. per channel requires.
An option is to use 15 nF. polystyrene caps. to connect the volume pot. wipers to the driver board. That gives you a high pass filter with an approx. 22.3 Hz. 3 dB. down point. Using the high pass filter is insurance against O/P trafo core saturation due to an excessive NFB low freq. error correction signal.
Use a separate bias pot. and cathode test resistor for each tube. The tubes available tend not to be as tightly matched as 1 bias pot. per channel requires.
An option is to use 15 nF. polystyrene caps. to connect the volume pot. wipers to the driver board. That gives you a high pass filter with an approx. 22.3 Hz. 3 dB. down point. Using the high pass filter is insurance against O/P trafo core saturation due to an excessive NFB low freq. error correction signal.
MAYBE, if the pair of EL34s in each channel were perfectly matched. Since the matching is NOT perfect and the tubes will age at different rates, separate bias adjustments for each O/P tube are very much in order.
Loop NFB, which the ST70 uses, has a definite homogenizing effect. 716P Series Orange Drops are quite adequate for this job. Save the "boutique" stuff for an amp that has zero loop NFB. In that situation, you might hear the difference between dielectric and/or plate materials.
labjr said:
You can run some of the ST-70 driver boards open loop. For instance, the DiyTube manual has a couple measurements of it running open loop. The Curcio board may do even better, but measurements are unavailable. In general, a small amount of NFB (10-20dB) is not a terrible thing in a push-pull power amp.
There is an alternative design that uses EL34 / KT88 tubes (triode strapped) that's popular around here called the PP-1C or PP-2C. It doesn't use negative feedback. Power output with the KT88s is roughly 20 watts, if I remember right, though you may prefer to go with a bit less power and the EL34s if you prefer that tube. Do a search for PP-1C or PP-2C. The full PP-2C circuit is not public domain, but if you email Allen Wright he'll give you a copy, or at least he used to do this.
BTW, the audibility of caps is a perennial topic of contention around here. Personally I believe they can make a significant difference, particularly electrolytics, but proper bypassing (parallelling an electrolytic with a small, high-quality film cap) is more effective than buying boutique caps. As good as they are, you still need to bypass Black Gates/Elna Silmics anyway.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.