DIY Audio Projects Forum

No comments? Somebody out there must have an opinion on whether this will work and why or why not!

I'm convinced it will work unless someone can show me where I've driven into a ditch.

Here's another schematic of the already tested (successfully) CCS. Also shown are the +500V and -500V power supply connections for the GM70 power stage. If I get no comments telling me why this won't work I'll go ahead and build it. Given I'm in the process of looking for a new house, renting my existing one, moving my belongings into storage, attemting to write and record some original music, and preparing for the arrival of my first child (gonna be a dad in about 6 or 7 months!) this build may not be completed for another couple of months but rest assured you will see pictures of the finished amp. Whether it works or not is another story. With all the knowledgable people here on this forum and the help I've received from you guys I think it WILL work.

As always any comments or criticism is welcome and appreciated.

_________________
Stephen

Stephen;

Looking at your suggestion it seems to me that in theory the circuit should work fine. Additionally, it should cut the stress on the O/P transformer in half just as you presume.

However, my concerns with this design are about the implications for the physical build rather then the circuit theory itself. My real concerns stem from the fact that your entire driver stage cathode circuit is sitting approximately 500v below ground.

#1 - The LM317HV and the BUX85 will both need to be heat sinked. Given the electrical isolation characteristics of the available packages it will be impossible to ground the heat sinks due to internal device electrical stress. As such, your design will have large pieces of metal inside the chassis at potentially lethal voltages with respect to ground. This presents a hazard to you as to service the equipment. Very careful layout and construction techniques will be required. As well as documented service procedures for the amp.

#2 - The DC heaters for the GM70s will also be almost 500v (minus bias voltage) below ground. As the entire DC circuit will be at this potential is is imperative that you make sure that any transformer you use to generate these voltages have appropriately high isolation to core and case. It is my experience that most small low voltage transformers lack the voltage isolation required.

#3 - The bias supply for the BUX85 also has the same considerations as the heater supplies.

#4 - My last concern is about the coupling to the driving stage. I see that you have stipulated the coupling caps have at least a 1000v rating, but that might not be enough. The leakage specs for the capacitors at this high voltage may be such that you still have a small DC current flowing into your grid circuits. With 150k grid resistors this could easily wreak havoc with your grid bias voltages. It could even potentially be enough to drive the grids positive and the tube into thermal runaway. You will have to be very careful how you pick those coupling caps. Pre assembly testing for leakage at high voltages is a must.

In short, although it is an interesting approach to the output transformer isolation problem, it may be simpler to design a more conventional stage and just go with an appropriately tested HIPOT output transformer. For example, Hammond states right in the catalog that their single ended output transformers (1630SEA etc) are HIPOT tested to 2000VRMS. They don't say what voltage the PP transformers are good to but I would suspect it's along the same lines. A quick call to their Engineering department would certainly answer the question.

It's not that I think the design won't work (because I think that it will), it's just that it appears fairly high risk from the design and construction perspective. Just my thoughts on the subject.

_________________
Matt
It's all about the Glass!
http://www.CascadeTubes.com
Cascade Tubes Blog

Good point. The plan was to use mica insulators and the chassis as the heat sink. This amp will be fan cooled by necessity - no way around it. But I can make the fan very quiet. I know I'm dealing with some serious voltages here. The construcion techniques will emphasize safety.

I'm using SMPS (switched mode) for the heaters. The isolation voltage between the primary and secondary should be OK - I checked - but you're right about many small trannys not being able to withstand these kinds of voltages.

Agreed. This is one I'll need to test before I put it in the amp. I have small low voltage DC power supply with very low current out to bias the BUX85. I'll put it through an isolation test before I use it. If it fails, I can always put a few extra well insulated windings around my toroidal power tranny to get a few volts DC with extremely high isolation!

This didn't occur to me. I was going to use PIOs. I don't know if PIOs have a low enough leakage or not. PI'll put 600 volts DC across the coupling cap and measure the leakage current per your suggestion. If it leaks enough DC through the 150K grid resistors to drive the grids high I need to find another cap with less leakage - either a different type or a higher voltage cap.

Thanks for the input. That's exactly what I wanted to hear. The coupling cap leakage due to the high voltage difference was something I missed. Otherwise you've confirmed my faith in the design, with adequate safety precautions in place.

Edcor only hi-pots to 1000VRMS. I believe Hammond hi-pots both their PP and SE OTs to 2000VRMS. Don't know why Edcor goes so low. As I recall I had a hard time finding Hammonds of the right impedance.

BTW, the design will use a LTP with an active CCS to split the phase. I have now what I think will be the final circuit. The op-amps are gone. I'll post as soon as it's drawn.

Construction set to resume...??? I'm ready to go, it's just a matter of making time now. Thanks very much for the input.

_________________
Stephen

Regarding the hipot test for Edcor, don't they use AC voltage on the test? 1000V AC RMS is 2828V AC peak to peak.

I would think 1000V DC would be fine on them, or am I missing something?

I agree that 1000VDC would be fine, but remember that's superimposed on the AC signal, and the resulting peak voltage could exceed the hipot test.

Consider this. For a tranny with a 10K primary (like mine) it will take 1000VRMS across the primary to achieve the 100 watt rating of the transformer - the same voltage at which it was hipotted. Based on that, and based on the fact (I'm reasonably certain) that Hammond uses 2000VRMS, IMO there is cause for concern. I'll have potentially 2828V AC peak added to the bias voltage. The combination of the two voltages could exceed the voltage used to hipot. IMO, based on my experience, the hipot voltage should be well above the voltage the transformer is ever expected to experience under normal operation. Even neglecting the DC bias in this case, the hipot voltage seems precariously close to the actual operating voltage.

_________________
Stephen

The Mu of the GM70 is about 7. So your driver would need to swing 400 volts P to P to get a 2828V P to P signal voltage, right?

It's a push-pull amp, not an SE, so it would only take half of the 400 volt swing on the drivers to get 2828V p-p out. I'm shooting for about 250 volts peak to peak output from the drivers so the power stage saturates before the driver.

I was merely pointing out that the transformer is rated for 100 watts and that it would take 1000VRMS - the same voltage used to hipot - to get that amount of power. I just think that's too low of a voltage to hipot a 100 watt 10K tranny that needs 1000VRMS to operate within its specified capabilities. If the transformer barely passes the hipot test and the insulation degrades slightly over time you might have problems down the road.

2000VRMS used by Hammond sounds a little safer to me. Any borderline devices will be destroyed by the test and culled before the product gets shipped to be used in someones home.

_________________
Stephen

Don't know how accurate this is, but it's a PSPICE model of the GM70 I found that was only very recently posted - I've been looking for months:

.SUBCKT GM70 A G K
+PARAMS: MU=8.037 ERP=1.5
+ KK1=4121 KP=182.25 KVB=34 vg0=-5.7
+ CGA=12.p CGK=8.p CAK=4.p RGI=1000

E1 7 0 Value = {(V(A,K)/KP)*LOG((1+ EXP((KP/MU) + ((KP*(V(G,K)+.5))/(SQRT(KVB+V(A,K)*V(A,K)))))))}
G1 A K VALUE = {(((V(7))^ERP)/KK1)*(1+SGN(V(7)))}

RE1 7 0 1e12

D3 5 k DX ; FOR GRID CURRENT
R1 g 5 {RGI} ; FOR GRID CURRENT
Rak A K 1G
Rgk G K 1G

C1 G K {CGK}
C2 G A {CGA}
C3 A K {CAK}
.MODEL DX D(IS=1N RS=1)
.ENDS

It appears to work. I can now model most of the GM70 amp, which incidentally has had some more changes made since the last schematic.

I'm also starting to get a handle on how to create custom models by starting with a similar tube and modifying the parameters within the model editor. Maybe one day I'll be even be able to create OT models of my own.

_________________
Stephen

Believe it or not I still have not decided on a driver tube for this project. I'm seriously starting to consider a step up interstage transformer so the GM70s get enough drive voltage. I can't find a driver tube that can swing 240 volts peak to peak. I've had several suggestions but none will output a clean 240 volt p-p signal. I'm having a lot of trouble finding a low power, high voltage tube.

I came up with an SRPP using 6SN7s that will do the job. However it will require about 600 volts on B+ to output a clean 240 volt peak to peak signal! It appears to me that it will work. None of the tube ratings are being exceeded under steady state operation. It's startup and shut down I'm concerned about. However, by taking a tip from another circuit (the Tim Mellow 6C33C OTL) and using neon lamps to shunt any high startup voltages, or if necessary even gas discharge tubes, I should be able to construct the circuit such that it can safely start and shutdown without any issues. The 600 volt supply will be readily available from the +650 volt rail to ground. (The +/-650V supply I’m using is equivalent to 1300V on B+).

Has anyone ever tried using a very high voltage on an SRPP driver to achieve very high output voltage swings, and if so did you use any form of circuit protection?

One last thing I’d like to point out is that the SPRR using 6SN7s with a 600volt B+ will use TWO tubes, NOT two triodes that share a heater. The ONLY way this will work is to bias the heaters of each triode so that the heater-cathode voltage ratings are not exceeded. So I’ll need to splurge for two extra tubes and sockets compared to using a single 6SN7 for each of the SRPPs. If I can get the circuit to start and turn off without issues, it will work. I could just slap MOVs everywhere I suspect there may be a problem, but I’d rather only use a lower capacitance device and then only where absolutely needed. PSPICE won’t simulate start up voltages by itself. You can however simulate them by making the tubes non-conductive and applying B+. I don’t know if that is adequate or not, but somehow I suspect not. I haven’t given it enough thought yet, but I’ve created the circuit I want to use with all the component values required to bias the tubes at about 10ma and swing an clean output of 250 volts p-p into a 10K load. Raising the load resistance reduces the distortion according to the simulation, but I wanted to assume worst case. I can’t imagine that the grid of the GM70 presents an impedance nearly that low. My guess is the distortion is very low since it will swing over 300 volts before any perceptible difference in the waveform can be seen (when the output is superimposed on a pure sine wave for reference), and if the load impedance is raised it even looks clean at 300 volts peak to peak. That’s more than enough to drive the GM70 to full power and clipping of the output stage.

If anyone has any last minute suggestions for driver tubes I'm all ears. Any suggestions would be appreciated. It's close to build time. The chassis has been drilled and painted, and I have all the parts including interstage trannys if I decide to use them to step up the voltage. It's decision time. I have 3 distinct topologies for the power stage and 3 for the driver stage, and I'm sill at a loss as to which ones I should build.

_________________
Stephen

I'd go with the SRPP driver. This is the job for which it was designed originally (large voltage swings). However, a µ-follower would provide lower distortion. Since it's a voltage driver, the minor efficiency loss of the µ-follower shouldn't be a problem.

I'd also stay away from the interstage transformer. It looks like an easy solution but for the levels for which you are looking you'll need a power driver stage; maybe a 6L6 (6V6 might work depending on the windings ratio). The problems associated with cascading two power amplifiers are hugh compared to the relatively simple matter of providing simple voltage amplification. Just my worth.

_________________
Matt
It's all about the Glass!
http://www.CascadeTubes.com
Cascade Tubes Blog