what are your feelings about chinese 5AR4 tubes

[gadget73]

Gadget, there's "works" and then there's .... "...WhyTF is it making that noise... I just can't seem to find the problem... why is that voltage off from... " So would anyone stick a 50R or more resistor under their filters? The chassis is just a wire to the transformer. What's the internal resistance of a rectifier? I didn't say it wouldn't work, it does, but this isn't about creating a bias tap for cutting cost. I'm just saying it's better engineering to not have any resistance between the filter (-) and the PT CT. Wanna put 10R under your filters for some cool reason.. go ahead.. or maybe buy a cap that has .1 or .2R more ESR that some other cap...Oh, my! No, No..! You know .... go ahead and put 20R on the CT for back bias, but don't consider puting 5R in the primary to lower all voltages.

.

If you can demonstrate that this performs worse than a standard tube rectifier used in the usual way I'll listen. In the meantime let's dispense with the histrionics. Electrically this is not different than a plate in each end with a grounded center tap. Does it make for the stiffest possible supply? Certainly not, but tube rectifiers never will. They are by their nature a resistance in the current path.

This is not back bias just to be clear but functionally it is very similar so it's not like this is a completely unhinged idea from the depths of my insanity.


As to that resistor in the primary thing I did prove my point with data. Its also not relevant to this.

 

[AmadeusMozart]

Does it make for the stiffest possible supply? Certainly not, but tube rectifiers never will. They are by their nature a resistance in the current path.

Up to a certain point the stiffness of a power supply is rather irrelvant when having an output stage running in class A. Especially when the last capacitors (CLCRC for the output stage going indivdually to the SRPP drivers using another RC) are of a decent capacity.

My amplifier has some 2W 100 Ohm spectrol adjustable resistors for adjusting cathode feedback in series with a 100 ohm resistor that has some sockets on the chassis to measure the current. Output transformers are ISO FC-20S, mains transformer is a Hammond 372FX with different bell covers so it can be mounted flat on silcicon rubber grommets (avoiding mechanical chassis noise) and a Hammond 5H choke (193H). Full set of pictures are somewhere in recent postings.

I do not know what it is with cathode feedback but I like it better than anything I've had previously.

 

[TriodeLuvr]

Up to a certain point the stiffness of a power supply is rather irrelvant when having an output stage running in class A. Especially when the last capacitors (CLCRC for the output stage going indivdually to the SRPP drivers using another RC) are of a decent capacity.

Are you aware the SRPP stage above isn't push-pull when it's loaded this lightly? The currents through the two tubes are actually in phase. That means there's no harmonic cancellation, and in addition, because both tubes share the B+ potential, maximum voltage swing (headroom) is likely to be less than when a single triode is used. Gain is also substantially less than a single triode IIRC. The only good use for a stage like this is when A) load resistance is sufficiently low to force push-pull action or B) a very large swing is needed AND a very high B+ can be applied. The SRPP stage used here is actually just a totem pole, and it offers none of the possible advantages that this circuit is capable of providing.

Jack

 

[AmadeusMozart]

Are you aware the SRPP stage above isn't push-pull when it's loaded this lightly? The currents through the two tubes are actually in phase. That means there's no harmonic cancellation, and in addition, because both tubes share the B+ potential, maximum voltage swing (headroom) is likely to be less than when a single triode is used. Gain is also substantially less than a single triode IIRC. The only good use for a stage like this is when A) load resistance is sufficiently low to force push-pull action or B) a very large swing is needed AND a very high B+ can be applied. The SRPP stage used here is actually just a totem pole, and it offers none of the possible advantages that this circuit is capable of providing.

Jack

SRPP has very little distortion on large swings and it is virtually immune to load changes. The top tube works similar to a CCS but I do not want to use semiconductors (except rectifiers in the PSU). After modelling all kind of different topologies with different tubes this one offered the least overall distortion. The downside is the Vhk limit and the filament has to be raised. The other item to keep in mind is the maximum cathode-filament resistance, for "normal" operation 20K but in phase splitting duties it may be 150K so the voltage divider from B+ has to take that into account. And yes, I'm very aware that gain is about half of what a single triode is rated at. But at the end of the day it is the synergy of the whole, not the performance of anything in isolation. The distortion results are very respectable, it's even better than when GNFB was to be used. And the 20V plate-screen difference is optimum - originally I looked at 33V as per Pearl-Hifi.com but that resulted in more distortion, same as when screen was at the same level as the plate.

Modelling was done for UL+GNFB, UL +GNFB +CFB, UL+CFB, pentode + GNFB +CFB, pentode+ GNFB, pentode + CFB, and all the previous with different voltage drops between plate and screen. Very time consuming which was OK because I was waiting for parts to arrive....(took me almost a year to find everything I wanted)

I modelled not only at maximum power but also at lower powers, all the way down to the mW region and it showed some very interesting differences. Test results with my HP 339A and with a Focusrite Solo running REW confirmed the modelling - I' was pleasantly surprised with the end results being close to the modelling.

YMMV

 

[TriodeLuvr]

The top tube does not function as a CCS. Broskie goes into some detail on this particular aspect of the circuit in Tube CAD Journal here (top of second column):

https://www.tubecad.com/articles_2002/SRPP_Deconstructed/page4.html

I analyzed and measured the SRPP in great detail some years ago. That was long before I read any of Brokie's papers, and I came to the same conclusions regarding fundamental behavior. If you like it, you like it, but its technical characteristics are as I described.

Jack

 

[AmadeusMozart]

The top tube does not function as a CCS. Broskie goes into some detail on this particular aspect of the circuit in Tube CAD Journal here (top of second column):

https://www.tubecad.com/articles_2002/SRPP_Deconstructed/page4.html

I analyzed and measured the SRPP in great detail some years ago. That was long before I read any of Brokie's papers, and I came to the same conclusions regarding fundamental behavior. If you like it, you like it, but its technical characteristics are as I described.

Jack

You may have missed a crucial word in "The top tube works similar to a CCS" and the paragraph prior to what you related mentions "then the single-ended/constant current source interpretation is half correct".

In the end the semantics are not important, one can argue about till the cows come home. What is important that this solution in the end had the best performance from all the solutions I modelled and that was proven in implentation.

If you can some up with something that would work better then I like to know. (I likely spend the better of 9 months time on modelling this design in all its variations and tested more things than I prefer to remember).

Peace.

 

[gadget73]

Looks like a mu follower, which is sorta-kinda just a triode with a cathode follower coupled to it.

Up to a certain point the stiffness of a power supply is rather irrelvant when having an output stage running in class A.

very true. Matters more in something like an AB amp with wide swings in current load though. Way back a bunch of pages ago this got started specifically about an ST-70 which is not a Class A setup.

 

[AmadeusMozart]

very true. Matters more in something like an AB amp with wide swings in current load though. Way back a bunch of pages ago this got started specifically about an ST-70 which is not a Class A setup.

That certainly does change things somewhat. But then what is the timeframe of these wild swings happening? When enough capacitance after the choke then to a large extend the fluctuations can be relatively mild. After all the DC resistance of the transformer windings is higher (in my case double) than that of the damper diode. We should not forget that the voltage drop across the damping diode is not linear when current changes (unlike a resistor).

Back to SRPP, or whatever name someone wants to give it: What attracted me to it was the writeup by Merlin Berclowe in the document "The optimised SRPP amp (part 1)" where he stated that "Consequently, this circuit - which we might call a 'half-mu' is very usefull indeed, since the gain and Miller capacitance remain quite constant despite aging of the valves, or the use of valves of varying provenance, because u is the most consistent and stable parameter of a triode. It should also be quite immune to changes in heater power and supply voltage. On the other hand, the main disadvantage is that the PSRR is only 6dB." (and imho also the need to elevating the filament supply in order not to exceed Vhk)

Since I orginally started out with wanting no global negative feedback and only cathode feedback the amplification stability of the driver stage is rather important to me. I later explored alternative designs, with using GNFB, UL, dropping screen voltage in regards with the plate all thrown into the mix, yet in the end I came back to where I started from.

 

[gadget73]

true, voltage drop is not linear on a damper, but its not linear on a standard full wave rectifier either. Functionally speaking, trading something like a 5AR4 at the ends of the winding for a reasonably stout damper at the center wouldn't really change the situation. It would just change the specific amount of effective resistance added to the circuit from whatever the 5AR4 is adding to whatever the damper adds. I expect most of them would be less drop than the typical full wave rectifier. The down side is just the practical aspect of using one. A lot of the high current dampers use goofy sockets and run on 6 volts, so not exactly a drop-in for an amp with an octal socket and a 5v heater supply.

all that is less an issue on a from-scratch design though.

 

[maxhifi]

That certainly does change things somewhat. But then what is the timeframe of these wild swings happening? When enough capacitance after the choke then to a large extend the fluctuations can be relatively mild. After all the DC resistance of the transformer windings is higher (in my case double) than that of the damper diode. We should not forget that the voltage drop across the damping diode is not linear when current changes (unlike a resistor).

Back to SRPP, or whatever name someone wants to give it: What attracted me to it was the writeup by Merlin Berclowe in the document "The optimised SRPP amp (part 1)" where he stated that "Consequently, this circuit - which we might call a 'half-mu' is very usefull indeed, since the gain and Miller capacitance remain quite constant despite aging of the valves, or the use of valves of varying provenance, because u is the most consistent and stable parameter of a triode. It should also be quite immune to changes in heater power and supply voltage. On the other hand, the main disadvantage is that the PSRR is only 6dB." (and imho also the need to elevating the filament supply in order not to exceed Vhk)

Since I orginally started out with wanting no global negative feedback and only cathode feedback the amplification stability of the driver stage is rather important to me. I later explored alternative designs, with using GNFB, UL, dropping screen voltage in regards with the plate all thrown into the mix, yet in the end I came back to where I started from.

I like this design, it has some elegance to it. Thanks for posting it!

 

[Mr Boochie]

That certainly does change things somewhat. But then what is the timeframe of these wild swings happening? When enough capacitance after the choke then to a large extend the fluctuations can be relatively mild. After all the DC resistance of the transformer windings is higher (in my case double) than that of the damper diode. We should not forget that the voltage drop across the damping diode is not linear when current changes (unlike a resistor).

Back to SRPP, or whatever name someone wants to give it: What attracted me to it was the writeup by Merlin Berclowe in the document "The optimised SRPP amp (part 1)" where he stated that "Consequently, this circuit - which we might call a 'half-mu' is very usefull indeed, since the gain and Miller capacitance remain quite constant despite aging of the valves, or the use of valves of varying provenance, because u is the most consistent and stable parameter of a triode. It should also be quite immune to changes in heater power and supply voltage. On the other hand, the main disadvantage is that the PSRR is only 6dB." (and imho also the need to elevating the filament supply in order not to exceed Vhk)

Since I orginally started out with wanting no global negative feedback and only cathode feedback the amplification stability of the driver stage is rather important to me. I later explored alternative designs, with using GNFB, UL, dropping screen voltage in regards with the plate all thrown into the mix, yet in the end I came back to where I started from.

I love the Valve Wizard, Blencowe's, writing style. I have read most of his books and postings. Odd, I have never seen a photo of him.

 

[TriodeLuvr]

Back to SRPP, or whatever name someone wants to give it: What attracted me to it was the writeup by Merlin Berclowe in the document "The optimised SRPP amp (part 1)" where he stated that "Consequently, this circuit - which we might call a 'half-mu' is very usefull indeed, since the gain and Miller capacitance remain quite constant despite aging of the valves, or the use of valves of varying provenance, because u is the most consistent and stable parameter of a triode. It should also be quite immune to changes in heater power and supply voltage. On the other hand, the main disadvantage is that the PSRR is only 6dB." (and imho also the need to elevating the filament supply in order not to exceed Vhk)

Since I orginally started out with wanting no global negative feedback and only cathode feedback the amplification stability of the driver stage is rather important to me. I later explored alternative designs, with using GNFB, UL, dropping screen voltage in regards with the plate all thrown into the mix, yet in the end I came back to where I started from.

Yes, Broskie says "half-correct," but he goes on to say in no uncertain terms that the constant current source aspect is false. I don't consider that to mean "similar." It either works (as a CCS) or it doesn't. Second, it's been a long time since I read the Merlin paper, but I don't believe his "optimum SRPP" is represented by the component values you're using. If I'm wrong about that, let me know.

A basic SPICE model driving a high-Z load indicates the SRPP to be inferior to a single-ended grounded cathode amplifier in terms of maximum voltage swing. Even at reduced output levels more suitable to its capabilities (and likely sufficient to fully drive a 7591), the best that might be said of the totem pole under these conditions is that it's pointless but does no harm. Actually, I'm not sure the latter is true, but I'll let it go.

Like Broskie says, if the SRPP isn't operating in push-pull, it's not SRPP. And SPICE makes it further evident that only in push-pull does the SRPP operate with less distortion. In light of this, plus the fact most hi-fi applications would load it incorrectly, resulting in single-ended operation, it's advantages are almost never realized.

I use the totem pole configuration in my 211 SETs, but only because the usual small signal tubes aren't rated for the anode voltage needed to create the necessary voltage swing. This is one of the few legitimate uses for this topology from a technical standpoint, other than driving coaxial cables or other low-Z loads. Again, if you like the way it sounds, that's fine, and my results indicate satisfactory performance as well. It has distinct limitations though, and a simple single-ended device can easily surpass it.

Jack

 

[AmadeusMozart]

Yes, Broskie s.....

Jack

It is Merlin Blencowe, not Broskie. <wink>

I remember that in days long past this SRPP, totem pole or whatever fancy name you want to give it, was a favoured high end solution. After all manufacturers would implement whatever was the cheapest, even if it saved them only the costs of one resistor if it had the same performance (or close to it).

The fact remains that modelling has shown that this overall topology is superior to all the other things that I modelled. And testing show very satisfactory results that actually exceed the modelling.

Perhaps if you look at a driver stage on its own (without taking in consideration the remainder of the circuit) there may be better solutions but it is the synergy between the topologies and tubes., i.e. the overall result, that counts in the end.

FWIW The late Patrick Turner was very fond of this topology as well. I've some fond memories of exchanging ideas with him.

You mention your 211 amplifier. I know the 211 will produce more output but what are its distortion figures at 1mW, 10mW and 1 Watt? After all it is the first watt that counts in a domestic environment.

Thanks.

Peace

 

[gadget73]

oddly enough a low-cost SRPP preamp has come to land on my bench just tonight. I'll map it out and see what its about. Power transformer is toast so it will be a little bit before I can power the thing up unless I want to drag out the old Heathkit HV bench supply. Parts count is quite low, looks like 2 caps and 5 resistors per channel but some of those may not exactly be part of the circuit.

also I'm just going by the fact the board says SRPP on it. I haven't actually determined what it is for sure.

also we've gotten just totally off topic here. This is from China, but it doesn't use a 5AR4. Sorry, best I can do on short notice

 

[Mr Boochie]

This is from China, but it doesn't use a 5AR4. Sorry, best I can do on short notice

I remember at one time, maybe 25 years ago when Mullard (Blackburn) GZ34 were starting to get scarce (read: super $$$) --- someone did a test on Russian Sovtek 5AR4 versus the Chinese Shuguang rectifier. The conclusion was the Russian 5AR4 put out more voltage, but not the current of the Shuguang tube. Anyway, I always buy my current batch of rectifiers from sellers that test the tubes and stand behind those tests. The factory QA has been notoriously inconsistent.

Any vintage amp above 20-watts per channel, I've gone to solid-state rectification. It's just more reliable.

 

[TriodeLuvr]

Jack

It is Merlin Blencowe, not Broskie. <wink>

I remember that in days long past this SRPP, totem pole or whatever fancy name you want to give it, was a favoured high end solution. After all manufacturers would implement whatever was the cheapest, even if it saved them only the costs of one resistor if it had the same performance (or close to it).

The fact remains that modelling has shown that this overall topology is superior to all the other things that I modelled. And testing show very satisfactory results that actually exceed the modelling.

Perhaps if you look at a driver stage on its own (without taking in consideration the remainder of the circuit) there may be better solutions but it is the synergy between the topologies and tubes., i.e. the overall result, that counts in the end.

FWIW The late Patrick Turner was very fond of this topology as well. I've some fond memories of exchanging ideas with him.

You mention your 211 amplifier. I know the 211 will produce more output but what are its distortion figures at 1mW, 10mW and 1 Watt? After all it is the first watt that counts in a domestic environment.

Thanks.

Peace

Well, I certainly do know about the first watt. My primary amp for more than 10 years has been a 4W/ch dual-mono SEP. Considering how I typically listen, I might even say it's about the first quarter-watt. My 211s have been in storage for years. They were originally built as prototypes, and initial measurements were made way back when. If they hadn't been very good, I wouldn't be spending time now on the rebuild and cleanup effort. Anyway, my experience modeling, testing and listening to garden variety SRPPs does not agree with yours. I guess we'll have to leave it at that.

Jack

 

[AmadeusMozart]

oddly enough a low-cost SRPP preamp has come to land on my bench just tonight. I'll map it out and see what its about. Power transformer is toast so it will be a little bit before I can power the thing up unless I want to drag out the old Heathkit HV bench supply. Parts count is quite low, looks like 2 caps and 5 resistors per channel but some of those may not exactly be part of the circuit.

also I'm just going by the fact the board says SRPP on it. I haven't actually determined what it is for sure.

also we've gotten just totally off topic here. This is from China, but it doesn't use a 5AR4. Sorry, best I can do on short notice

Yes, totally gone off topic. Read some book excerpt the other day about the end game in vacuum tube technology and how much testing was done, how critical the coating of the cathode was, the failure rates, the amount of labour involved. Unfortunately I deleted the link and the excerpt, I've been tidying up my PC now that I am calling it quits to making stuff after 57 years. Fortunately this last amplifier has proven to be more satisfactory (and far exceeded expectations) than anything else I've owned and listened to.

Regarding a SRPP preamplifer, this is what Patrick Turner wrote about one:
https://web.archive.org/web/2019040...dio.com.au/preamp-10tube-integrated-2006.html

And Jack:

Regarding a first quarter watt: Same here, more often than not, it hoovers around the 50mW to 100mW.

The selection of the two tubes working in synergy is important, e.g. an EL84 or EL34 did not perform as well in conjunction with the ECC81 / E81CC. And for cathode feedback to work well you need a tube with low drive requirements, KTxx and 6L6xx were disasters. (I started out with KT66, someone locally has still a genuine batch of NOS). Neither did one of the other well known double triodes work well, both the noval series and octal series.

Often people look at parts in isolation which imho is only going to give you a starting point. I had the time, it was something to keep me occupied, to go through all the different solutions. Don't kn ow what I'm going to do now to keep me occupied since health put a stop to tinkering.

It's just as well you have a different experience, if there are no differing opinions then there would be no progress and life would be pretty boring. I will be interested to see the test results of your 211 amplifier once you've finsihed with the resoration, I'll keep an eye out for it.

edit: link to the excerpt mentioned earlier: https://od.lk/s/OV8xMjUyMDMzMjVf/EndgameTubes.PDF
It is part of the book "Declining Demand, Divestiture, and Corporate Strategy By Kathryn Rudie Harrigan"

 

[TriodeLuvr]

My original measurements were taken before I had the current ARTA test system on my bench. The results were sufficient to know the amplifier was low distortion, including its operation into the A2 region. Nevertheless, I don't know the harmonic signature, so all new measurements are in order, and I'll post them here.

About the 5AR4, most of my higher power builds have used SS supplies. However, I built a 2X EL34 guitar amp for my daughter last year using a Sovtek 5AR4. So far, the rectifier is holding up fine. Sovtek 5AR4s are still available from a few sources at decent prices.

Jack

 

[Mr Boochie]

My original measurements were taken before I had the current ARTA test system on my bench. The results were sufficient to know the amplifier was low distortion, including its operation into the A2 region. Nevertheless, I don't know the harmonic signature, so all new measurements are in order, and I'll post them here.

About the 5AR4, most of my higher power builds have used SS supplies. However, I built a 2X EL34 guitar amp for my daughter last year using a Sovtek 5AR4. So far, the rectifier is holding up fine. Sovtek 5AR4s are still available from a few sources at decent prices.

Jack

Did you buy the Sovtek from a source that tests them? Like McShane?

Thanks!

 

[TriodeLuvr]

I bought mine from Antique Electronic Supply as part of a larger order. Jim is an excellent source too, assuming he still has them.

Jack