PT15 PSE design

What people are working on at the moment
Post Reply
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#1 PT15 PSE design

Post by JamesD »

One of the designs I considered for my next main system amp was a PT15 PSE amplifier with an EL802 driving an interstage 2:1 step down with separate secondary windings one for each PT15. Possible problem is driving the Millar capacitance of the EL802. Traditional design would put a cathode follower here so I have a solid state unity buffer that is complete overkill but very low distortion and a benign harmonic distortion spectrum.

I thought I would post it as I think it is interesting - I haven't done an indepth design analysis on anything except the ss buffer so the circuit doesn't have values - hope its interesting...
ssbuffer EL802 PT15 PSE.png
ciao

James
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#2 Re: PT15 PSE design

Post by ed »

Intrigued by the front end James. I don't fully follow it(yet), but if it's not a silly question, why do it long hand?
Elliot makes the argument that it takes loads more space and costs a lot more than the equivalent opamp. I'm sure I'm missing something obvious though.

btw there is a long thread on world designs from the early days which makes interesting reading. You and Steve Shiels were discussing at length the similarities between the PT15 and the PX4/25.
There's nowhere you can be that isn't where you're meant to be
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#3 Re: PT15 PSE design

Post by JamesD »

If doing it "long hand" was a criteria - why do anything other than Class D amps with mp3 based front ends :lol: We design and build boat anchors here considerations of space cost and heat don't apply :lol: :lol: :lol:

But a serious answer is that the buffer sounds better as it has very low distortion and a more benign distortion spectrum, wider bandwidth both open and closed loop, lower noise, etc. etc. Op amps are three stage devices so its not a short feedback loop around them with the attendant distortion complexity that limits their sonic potential...

The design is a Compound Feedback Pair otherwise known as a Sziklai pair loaded with a CCS. I like the CFP configuration as with suitable choice of transistor pairs and operating points it provides really low distortion with a triode like distortion spectrum i.e. mainly 2nd harmonic with higher orders decreasing in amplitude by order and typically with anything over 4th harmonic in the noise floor. It can be arranged to provide gain as well at the cost of higher distortion but still lower than alternative transistor pairs. The pair gain is very close to that of a Darlington configuration. BUT the main reason for using it is that it sounds better than other ss gain or buffer stages so I've use it a lot over the years since having it demonstrated to me in the late 70s. Drawback is that the damn things can burst into oscillation so in practice a small capacitor between collector and base of the first transistor is sensible - around 30pF is usually enough.

So its a bit different and that was enough to encourage me to use it in the design - op amps are boring and I use the CFP in one of my latest RIAA designs so I thought I would use it here :-)

I must look up that thread again - thanks for reminding me of it.
User avatar
Paul Barker
Social Sevices have been notified
Posts: 8867
Joined: Mon May 21, 2007 9:42 pm

#4 Re: PT15 PSE design

Post by Paul Barker »

Would you share your’ buffer design? I’m messing about in the dark with that, your’e view on your’s sounds interesting.

Sorry about the problem in punctuation I cant make the various yours to look right but don’t know what to do about it.

I do have some Pt15’s but no idea where right now. But I’m busy enough planning whta I’m doing.
"Two things are infinite, the universe and human stupidity, and I am not yet completely sure about the universe." – Albert Einstein
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#5 Re: PT15 PSE design

Post by ed »

JamesD wrote: Thu May 14, 2020 6:03 pm

The design is a Compound Feedback Pair otherwise known as a Sziklai pair loaded with a CCS.
many thanks for pointing it out. Please spare a thought for the afflicted...I was messing with the very same thing about 18 months ago and I've got absolutely no memory of it whatsoever.....oh waily waily
There's nowhere you can be that isn't where you're meant to be
User avatar
izzy wizzy
Old Hand
Posts: 1496
Joined: Fri Nov 02, 2007 7:02 pm
Location: Auckland NZ
Contact:

#6 Re: PT15 PSE design

Post by izzy wizzy »

JamesD wrote: Thu May 14, 2020 4:18 pm One of the designs I considered for my next main system amp was a PT15 PSE amplifier with an EL802 driving an interstage 2:1 step down with separate secondary windings one for each PT15. Possible problem is driving the Millar capacitance of the EL802. Traditional design would put a cathode follower here so I have a solid state unity buffer that is complete overkill but very low distortion and a benign harmonic distortion spectrum.

I thought I would post it as I think it is interesting - I haven't done an indepth design analysis on anything except the ss buffer so the circuit doesn't have values - hope its interesting...

ssbuffer EL802 PT15 PSE.png

ciao

James
I gotta ask why not pp on the finals? You have everything there and pse makes the opt harder in the light of someone, could be you on another thread, that the pt15 requires a ton of inductance for good bass. PP OPT does that for free. I think you might have referenced poinz's music machine elsewhere; not too different.

Cheers,
Stephen
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#7 Re: PT15 PSE design

Post by JamesD »

Paul,
Sure thing - I'll dig it out and write some words for it. If you have a spec for what you want it to do I can optimise the design in LTspice to suit your use including some gain if required. It can be scaled up to 100Watts and more output to so hit me with your needs and we'll meet them :-)

Ed.
I'll dig out some reference papers and post them or links to them over the next couple of days for you. Nothing like a bit of research to cloud the brain...

Stephen,
Your so right :-) And I have done that as a sketch but I have a slightly different PT15 PP design as I like to use the iron concertina as the phase splitter so it doesn't need the buffer but is a bit more complicated to optimise as a phase splitter/ driver design with an SE gain stage to give a Ralph style PP overall topology. I need to get that into a shape to upload...

I have a older design too that I haven't built and now I think is too complicated using partial feedback (JRB term). Here is that design - I'm not sure it can be easily stabilised...
iron Concertina 6N30p 3Cg PT15 v0.2.png
J.
User avatar
ed
retired
Posts: 5384
Joined: Thu Jun 21, 2007 4:01 pm
Location: yorkshire
Contact:

#8 Re: PT15 PSE design

Post by ed »

JamesD wrote: Thu May 14, 2020 7:49 pm
Ed.
I'll dig out some reference papers and post them or links to them over the next couple of days for you. Nothing like a bit of research to cloud the brain...
thanks James...

very frustrating to forget things so quickly...I had to go and rummage for notes...found 'cfp vs darlington' and a whole bunch of notes on cascodes.....If I go through the whole lot again I feel sure I will have forgotten again by next week....negative thinking I know, but maybe armed with this knowledge I can save myself some time....time to give up before I become really dangerous....
There's nowhere you can be that isn't where you're meant to be
steve s
Shed dweller
Posts: 2829
Joined: Wed May 30, 2007 6:19 pm
Location: east yorks

#10 Re: PT15 PSE design

Post by steve s »

Below is an interesting way of getting round the problems of using parallel valves by adjusting the heater voltage
This artical is from the 50s but the issue was know about for a long time before that
I'm sure this method could be used with great effect to overcome the issues, which are usually very audiable.
Hopefully you can open it

I've copied it.. there is also an interesting artical on balancing push pull..
http://www.thevalvemuseum.org/sdhc/articles/art-248.htm
-------'------------------------------------------------'
Various methods for adjusting the balance of a pair of triodes are considered. A new method is suggested which is based upon the adjustment of the cathode temperature so that the static anode currents are made equal. Measurements indicate that under these conditions the mutual conductance are then nearly equal, and the behaviour with respect to overall heater voltage changes is also improved.
In many applications, and particularly in the case of DC amplifiers, it is desirable to have a pair of identical valves. For example, in push-pull power output circuits this is necessary to obtain an exactly balanced push-pull output. In the case of DC amplifiers the pair of valves are arranged so that the effect of changes in supply voltages on the characteristic of one valve are compensated by changes in a similar, preferably identical, valve. Similarly, compensation is obtained for any change in the valve characteristics with time. Unfortunately, it is very difficult to obtain identical valves, particularly in the case of valves with an oxide cathode; and some means must be adopted in the associated circuits of adjusting for differences between pairs of valves.

Triode Parameters
The small signal parameters of a valve (for the sake of simplicity a triode is chosen, but similar arguments apply to pentodes at fixed screen voltage) which are of interest are the amplification factor (μ), the mutual conductance (gm), and the static anode current (Ia), all measured under fixed conditions, ie, heater voltage, anode voltage, and grid bias. (Note the dynamic anode resistance is μ/gm and is hence implied from a knowledge of it and gm). The static anode current is important, as it determines the static DC balance of balanced circuits, obviously of extreme importance in the case of DC amplifiers, although it is desirable from many points of view that the anode currents be equal in push-pull AC circuits - ie, reduction of ripple, and of DC saturation of the output transformer.

If the parameters of a series of valves of the one type are measured it is found that μ shows small deviations of the order of a few per cent from the mean. The gm and Ia, on the other hand, deviate by well over ten per cent from the mean value. This is to be expected, as the amplification factor is, under ideal conditions, determined by the valve geometry, which with normal construction methods can be held to close tolerances, whereas the gm and Ia are also determined by the cathode emission, which is known to vary considerably in the case of oxide cathodes, even with the most carefully controlled production.

Balance Circuits

Fig. 1. Three basic methods used to balance the static properties of a pair of triodes.
In any double triode circuit it is possible to adjust for equality of either the static anode currents, the stage gain, or the effect of heater changes, but not for all three simultaneously. Normally, adjustments are made for equality of the static anode currents, or zero adjustment. Methods of doing this are shown in Fig. 1. For example, the anode supply may be taken through a potentiometer as in 1(a), and this adjusted for static balance. Alternatively, the bias can be changed as in 1(b) so as to give static balance. Finally, the valve may be shunted with a resistor as in 1(c). Unfortunately, if other adjustments are introduced so as to equalize also the stage gain and the effect of heater changes, the various adjustments are not independent, and in a multi-stage amplifier the use of three interacting controls on each stage would be impracticable.

More elaborate circuits have been devised to balance the anode current and the gain simultaneouslyb. However, these methods are based on the use of a large cathode resistance common to both valves, this serving to nearly balance the dynamic properties, and one of the methods of Fig. 1 is then used to balance the static currents. An example of this type of balance circuit is given in Fig. 2. However, this method is applicable only when using parallel balanced triodes with a large value of the common cathode resistor R, and hence is not applicable in many cases.


Fig. 2. A method for balancing both the static and dynamic properties of a pair of triodes.
New Balance Circuit

Fig. 3. New circuit for the overall balance of a pair of triodes
As the differences between triodes of the same type are mainly caused by differences in the emission from the cathode, it should be possible to balance valves by making the cathode emission the same. A new circuit4 based upon this fact is shown in Fig. 3. A potentiometer of a few ohms resistance is connected as shown so as to introduce a small difference ΔVh in the heater voltages of the valves 1 and 2. The simplest procedure is to adjust R so that the static anode currents of the valves are equal. If necessary, the overall heater supply can be increased slightly, but in many applications, for example in the case of DC amplifiers, it is desirable to run the heaters at a reduced voltage and the overall drop in R is not important.

It was found experimentally that the maximum difference in voltage between the two triodes amounted to 10 per cent of the normal for a batch of 50 twin triodes type 12AX7, and that of these approximately half could be balanced with heater voltage differences of only 5 per cent.

As the potentiometer R, is set so as to equalize the static anode currents, the valves will be perfectly balanced if the mutual conductances are equal and they respond equally to the effect of heater supply changes.

Table 1

Differences Between Twin Triodes Before and After Balancing by Heater Compensation.
These two points were checked by measurements on a number of valves type 12AX7. Measurements on the mutual conductance are given in Table 1. The twin triode type l2AX7 was first tested at fixed heater voltage, then the heater voltages were adjusted for static balance at the recommended test conditions (anode voltage 250 V, grid bias -2.0 V). The dynamic balance was next measured by measuring the mutual conductance, and also the anode current at a grid bias of -2.2 V and -1.8 V. It is obvious that for all the valves tested the mutual conductances after balancing for Ia are almost identical. This would probably hold only for triodes of near-identical geometry, where the unbalance is caused solely by the differences in work function of the cathode. Once the emissions are made equal by adjustment of the cathode temperatures, the valves should then be identical apart from a slight difference in the escape voltage of the electrons from the cathodes. It is possible that the differences in temperature will introduce other side effects such as differences in life of the two triodes, different rates of poisoning, etc. However, balance adjustments are normally a routine adjustment in the case of DC amplifiers. The twin triode No. 1 of Table 1 was tested after 100 hours and the balance adjustment was found to have changed by 0.07 V, the new value of ΔVh being 0.06 V. It should be noted that out of the nine valves tested, one valve (No. 6) showed anode currents and mutual conductances to be nearly balanced without any further compensation.

Of major importance in DC amplifier design is the elfect of heater supply variations, and several circuits are used to balance a pair of triodes for equality in changes due to heater supply changes2,3. Although the new circuit makes the cathode temperatures different, all other properties are nearly balanced, and to a first order it would be expected that the stability to heater supply variations would be also balanced.

Table 2

Measurements in Table 2 give the effect of a change of + and - 5 per cent on the overall heater supply. As is clear from these measurements, the compensation is good but not perfect. However, this result is typical of all heater compensation circuits, which rarely give an improvement of better than 20 :1 (Verhagen1).

Conclusions
It is concluded that the new balance circuit of Fig. 3 is capable of giving, simultaneously, an accurate balance of the static anode currents and mutual conductances of pairs of triodes. As the amplification factors are normally also balanced, this gives a pair of triodes, without selection, identical to within a few per cent. At the same time the balance with respect to overall heater supply changes is improved sufficiently to render heater compensation circuits unnecessary.

References
Last edited by steve s on Fri May 15, 2020 2:20 pm, edited 1 time in total.
The tube manual is quite like a telephone book. The number of it perfect. It is useful to make it possible to speak with a girl. But we can't see her beautiful face from the telephone number
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#11 Re: PT15 PSE design

Post by JamesD »

Interesting! I can access it so I'm joining as a friend of the museum and should get access in a couple of days.

thanks

James
User avatar
Nick
Site Admin
Posts: 15707
Joined: Sun May 06, 2007 10:20 am
Location: West Yorkshire

#12 Re: PT15 PSE design

Post by Nick »

"1950's" and "DC amplifier design" likely driven by military servo control requirements I would guess and balancing the LTP in op amps.

Not that that stops it being a interesting idea.

Is it this doc?

http://www.r-type.org/articles/art-248.htm
Whenever an honest man discovers that he's mistaken, he will either cease to be mistaken or he will cease to be honest.
JamesD
Old Hand
Posts: 997
Joined: Tue Oct 09, 2007 4:26 pm
Location: North Yorkshire

#13 Re: PT15 PSE design

Post by JamesD »

Fascinating that it works for IHT as well as it would for DHT! I know emission rate varies with temperature for both but, in general, its a much lower dependency for IHTs.

The more I think about this the more I like it - I was taught to always run a PP pair with 5% or so imbalance between the the anode currents and providing the OPT allowed for this then it would sound better as not all of the 2nd harmonic would cancel and this would balance the sound more with the third harmonic... This would be a neat way to do that to - I was taught to do it by a variation of circuit 1.b. in the article...

Need to do this in my next PP or PSE amp.

thanks

James
Post Reply