IME, the Russian/Chinese tubes match their own datasheets pretty well, but not if the tubes are re-branded/re-labelled and sold as some popular western "audio tubes", since they are different tube types to begin with.
Yes, that's my finding too.
So the 6П14П (6P14P), which was catalogued in Russia as an "Analogue" to the EL84 is differently designed, and measures somewhat differently.
If you buy a big lot of these and measure, the variations can be wide, too. As Anatoliy said - "like from Moscow to Vladivostok". (Where is Anatoliy (Wavebourn) these days)?
An Electroharmonix or Sovtek EL84 is a 6П14П, so it's best to use the Russian data sheet.
Similarly, the Sovtek 5881WXT is a 6П3C-E (6P3S-E), rather than a real 5881, and has wide variations too.
But wide variation does not mean they are weak or short-lived. I find them to be very tough, even when excessive voltages have been applied - as in guitar amps. Want to see >400V on a 6П14П/EL84? Look inside a Peavey Classic-50. (Yes, I measured it, too) Fixed bias, too.
All of these considerations emphasise the point that one must build and measure, and not base one's expectations on what data sheets say.
Last edited:
Joe are you coming to ETF this year? The preamp power supply shootout might help you find the diamond ;-)
Jan
In my post 154 I calculated
I will assume the same Rk for both valves. I will further assume that Rk is approximately 1/gm - it will almost certainly be within a factor of 2 of this.
One way to calculate the output impedance of a circuit is to find the unloaded (open-circuit output) voltage gain, and then divide this by the closed-circuit transconductance. This is exactly what we are doing when we use ra=mu/gm for a valve, but the principle applies to circuits too.
Consider first the balanced SRPP, with unbypassed lower cathode. With Rk=1/gm the effective anode impedance r'a will be approximately equal to 2ra. Open circuit voltage gain will be just below mu/2, as we just have a slightly modified active load. Short circuit transconductance will have two contributions: gm/2 from the lower valve, and gm/2 x Rk x gm from the upper valve. The latter part comes from the lower valve working into an anode load of Rk to develop a signal voltage which then drives the upper valve. The lower cathode degeneration halves its gm. Transconductance = gm/2 x (1 + ERk x gm).
So we have Zout = (mu/2) / [ gm/2 x (1 + Rk x gm) ] = mu/[ 2 gm ] = ra/2.
For the bypassed lower cathode we have voltage gain = mu x 2/3. Transconductance = gm x (1 + Rk x gm). Hence Zout = (2 mu/3) / [ 2 gm ] = mu/3 gm = ra/3.
It can be seen that the above numbers fit the expected pattern.
for a 12AT7 (with ra=15k). Some may be surprised to see such high figures. Isn't the SRPP supposed to have a low output impedance? After all, the upper valve is a cathode follower, isn't it? Here is a rough calculation to show that Zout will be around ra/2 or ra/3, depending on whether the lower cathode is bypassed or not.DF96 said:
I will assume the same Rk for both valves. I will further assume that Rk is approximately 1/gm - it will almost certainly be within a factor of 2 of this.
One way to calculate the output impedance of a circuit is to find the unloaded (open-circuit output) voltage gain, and then divide this by the closed-circuit transconductance. This is exactly what we are doing when we use ra=mu/gm for a valve, but the principle applies to circuits too.
Consider first the balanced SRPP, with unbypassed lower cathode. With Rk=1/gm the effective anode impedance r'a will be approximately equal to 2ra. Open circuit voltage gain will be just below mu/2, as we just have a slightly modified active load. Short circuit transconductance will have two contributions: gm/2 from the lower valve, and gm/2 x Rk x gm from the upper valve. The latter part comes from the lower valve working into an anode load of Rk to develop a signal voltage which then drives the upper valve. The lower cathode degeneration halves its gm. Transconductance = gm/2 x (1 + ERk x gm).
So we have Zout = (mu/2) / [ gm/2 x (1 + Rk x gm) ] = mu/[ 2 gm ] = ra/2.
For the bypassed lower cathode we have voltage gain = mu x 2/3. Transconductance = gm x (1 + Rk x gm). Hence Zout = (2 mu/3) / [ 2 gm ] = mu/3 gm = ra/3.
It can be seen that the above numbers fit the expected pattern.
I did get a bunch of Sovtek 6SN7's in a lot of tubes a few years ago and they did work well dropped into a circuit that I made using old RCA smoked glass tubes.
The Sovtek 300B's, 6CA7's KT66, and 5881's don't work like their US or British counterparts. They do work, and in some cases handle more power than their US equivalents.
I haven't found any Russian EL84, 7189 or 7189A that can stand up to the abuse that I regularly feed to NOS 6BQ5's from GE or Sylvania. The JJ EL84 WILL eat everything I feed it if you get good ones. The JJ's or NOS 6BQ5's will work good at 430 volts B+ with cathode bias and a 325 volts screen supply. With a 6600 ohm load I get 25 to 30 watts output and nothing glows or blows up even in a guitar amp.
Unfortunately I have found some rather wide variations in US or Canadian made tubes. During the final years of vacuum tube production the major tube companies were stuffing the glass with whatever they had to fulfill military contracts. Usually this resulted in a bigger set of guts than spec, but I have found some 6CW5's with tiny internals that red plate at 8 watts. They were Canadian GE's.
The only requirement was that the tube supplied to the US government worked in the equipment called out in the purchase contract. These tubes have now hit the surplus market and have caused some havoc.
There is a thread currently running on the diyAudio forums concerning the 6HB6. There are several users reporting a very wide variation in idle current for the same set of conditions. The numbers range from 5 mA to 70 mA for the same grid, screen, and plate voltages. At least two users have supplied curves that don't come close to the published data.
I have learned to read the data sheets, but not to rely on them until I at least spot check a few tubes.
How, if you use a good linear resistor?
Do we really care so much about absolute voltage regulation or are we looking for the noise benefits? The latter, I think, although they are tightly related.
Yes indeed! However, even after all this discussion, I'm fully prepared for it to be a 6x4 with chokes!
Last edited:
I use voltage regulation of a number of reasons in my phono stage designs, the two primary reasons are broadband noise and ripple and for what I term LF baseline wander (which is what DF96 is referring to in his comments) for want of a better term.
I tend to use cascode error amps with zener references, and pentodes/tetrode pass tubes with relatively well filtered screen supplies. Regulation and output impedance are pretty decent as well, because of the low output z I generally don't need a lot of decoupling between stages which results in very nice behavior below the audio band.
I also generally use CLC input filters and tube rectifiers - the rectifiers delay B+ a little and reduce voltage stress on components during warm up.
The combination of all of these things give me supplies with ripple and noise levels at the output that challenges the limits of my test equipment.
http://www.diyaudio.com/forums/analogue-source/213769-muscovite-6s3p-tube-phonostage.html
http://www.diyaudio.com/forums/analogue-source/273817-muscovite-mini-iii-6n23p-phono-stage.html
http://www.diyaudio.com/forums/analogue-source/254641-muscovite-mini-ii-6n14p-phono-stage.html
http://www.diyaudio.com/forums/analogue-source/246756-muscovite-mini-6-9-6z9p-phono-stage.html
As always there are design tradeoffs, I use 12AX7A in cascode connection which is not an ideal application of the tube, but gets me sufficient gain capability for the supply voltages I generally use, it's relatively cheap, quiet, and a comfortable choice for those few who might choose to copy something I have designed. I am sure there are plenty of comments someone could make about my approach to things, there are many paths leading to a satisfactory result and in practice this approach has proven very flexible, a decent performer, and reliable over decades of use. (Approaching 3 decades of use now)
The Muscovite is used with Lundahl LL1941 amorphous core transformers and an SPU A95 LOMC.
Many of the more interesting phono stages here at diyA will be found in analog source if you do a search.
I tend to use cascode error amps with zener references, and pentodes/tetrode pass tubes with relatively well filtered screen supplies. Regulation and output impedance are pretty decent as well, because of the low output z I generally don't need a lot of decoupling between stages which results in very nice behavior below the audio band.
I also generally use CLC input filters and tube rectifiers - the rectifiers delay B+ a little and reduce voltage stress on components during warm up.
The combination of all of these things give me supplies with ripple and noise levels at the output that challenges the limits of my test equipment.
http://www.diyaudio.com/forums/analogue-source/213769-muscovite-6s3p-tube-phonostage.html
http://www.diyaudio.com/forums/analogue-source/273817-muscovite-mini-iii-6n23p-phono-stage.html
http://www.diyaudio.com/forums/analogue-source/254641-muscovite-mini-ii-6n14p-phono-stage.html
http://www.diyaudio.com/forums/analogue-source/246756-muscovite-mini-6-9-6z9p-phono-stage.html
As always there are design tradeoffs, I use 12AX7A in cascode connection which is not an ideal application of the tube, but gets me sufficient gain capability for the supply voltages I generally use, it's relatively cheap, quiet, and a comfortable choice for those few who might choose to copy something I have designed. I am sure there are plenty of comments someone could make about my approach to things, there are many paths leading to a satisfactory result and in practice this approach has proven very flexible, a decent performer, and reliable over decades of use. (Approaching 3 decades of use now)
The Muscovite is used with Lundahl LL1941 amorphous core transformers and an SPU A95 LOMC.
Many of the more interesting phono stages here at diyA will be found in analog source if you do a search.
Good term! I like the concept, but does such variation actually occur at audio or single digit frequencies and/or appear in the output of unregulated preamplifiers?
I'll have an output transformer in my line driver, so I doubt that subsonics will pass that bridge anyway.
I have a large stack of preamp designs that I am trying to work through and always looking at new ones. To me, the power supply is what I need to look at more intently, since I'm building a pre for MC input with very efficient speakers.
In the distant past, I have had better music results with passive CL supplies but I think the dialogue on regulators has deepened a lot with forums such as this one and I work with tube preamps built by a company I'm involved with that are highly regulated, ridiculously quiet, and extremely non-mechanical sounding. When you get zero noise and hiss with an ear against 115dB horns, that is saying something.
I think that on-board decoupling with MOS cap multipliers that present a controlled low impedance through the range is part of the success of these commercial designs.
I don't doubt that I was using bad regulator designs in the 80s and early 90s which led to my rejection of the approach, but on the other hand, my tolerance for noise, hum, hiss has decreased. A good heavily filtered passive supply sounds better than a bad regulated supply, which should come as no surprise, but I want to give my current project a leg up.
Noise is less important than PSRR and Zout when it comes to a listener's assessment of quality.
It's "illuminating" to look at the FFT of a tone coupled to a power supply rail -- say a few mV. You get a picture of Zout and THD created by the power supply.
It does appear on the output of both line stages and phono pre-amps with unregulated supplies and to a considerably lesser extent on units with regulated supplies using unregulated filament supplies to heat their error amplifier filaments. If you have solid state woofer amps or large tube amps with good LF response it can cause woofer breathing or in the case of tube amps LF stability issues. I still remember connecting one of my very early pre-amplifier designs to a pair of VTL300 and watching the woofers flap due to VLF output in the pre-amp. The pre-amplifier in question had a line stage that was flat to -1dB down to a couple of Hz, and a phono stage that still had appreciable gain well below 20Hz. Did not get to listen that day!
I use a variant of my old transformer coupled dht line stage design (now using 12s) in my system and I don't have issues with subsonic response since the transformer blocks any VLF wander that is present. I would not expect you to have an issue either. My system still has significant acoustical output below 30Hz to my surprise.
I use cascode front ends in all of my current phono stage designs because I discovered a few years back that the performance of MM and LOMC (driving step up transformers) was significantly impacted by the miller capacitance of D3A and other similar types I favored at the time. (HOMC work quite well in this scenario) The cascode offers pretty low input capacitance and lots of gain, the trade off of course is nearly non-existent PSRR - I have measured as little as 1dB of PSRR... lol My second stages generally have good PSRR since I employ hybrid or choke/tube based mu follower variants in most case.
I have had to work on very quiet power supplies as a result, in particular since I don't like the traditional decoupling approach in the supply rails. I should mention that in my designs the supply is definitely an audible part of the overall sonic equation.
I have had to work on very quiet power supplies as a result, in particular since I don't like the traditional decoupling approach in the supply rails. I should mention that in my designs the supply is definitely an audible part of the overall sonic equation.
I agree with that, with the addendum that it is a different game with ultra high sensitivity speakers. Trash that is inaudible directly on average box speakers becomes quite apparent with big horns. 120hz buzz kills me. I hate it.
Of course, there may be other artifacts introduced by PSU nasties which may be audible to some degree, as you intimate, but I'm speaking here of straight up loud identifiable noise.
Are you attributing this VLF signal to AC line variation? I'm surprised that line voltage varies that way but it might. I think DC pulses on the line, which are apparently quite common and introduced by asymmetrical loads somewhere on your utility branch could zap your transformer operation, especially with toroids, creating a similar effect.
And maybe 1hz response is just a bit excessive?
I have had LF instability/speaker pumping in amps but never blamed the AC line.
I have seen what looks like (temporary) DC level shfts on the output of my 5-20 amp. Subsonics can get through an OPT, even though they will be attenuated. The basic idea is that you don't want big signals you can't hear messing around with the bias of stages carrying signals you can hear. This includes the OPT and the speaker. A line stage transformer may pass subsonics even better, as it is likely to have a wider bandwidth than a typical OPT.Joe Roberts said:
It certainly does in the UK. I would be surprised if it did not do the same elsewhere. When I found this stuff on my amp output I was puzzled about where it came from, until I noticed it was correlated with shifts in the HT rail voltage - which can only come from mains variations.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.