GM70 Amplifier

[Paul Barker]

I’m in a fragile state of forsed full time carer situation. The PSE 801a amplifier has gone single channel, which Im not in a fit state to fault find.

So if I repurpose the power supply which is choke input valve rectified 630 v dc generated by a 625-0-625 @.5A transformer with alift from choke input by a small 1 microfarad first capacitor. The drop from a parallel pair of potato mashers with a current drawer on the cathodes of the 4 pcs 801a stles at the above 630v.

So the full power supply untouched should satisfy the current requirements of two gm70’s at 630v.

Because of my state of mind the signal section would be simple. New +/- power supply call it 200v C+200v/C-200v for an el84 direct coupled driver.

B+ 630v for the voltage amplifier power supply reduced appropriately.

So output stage A2.

I was getting 12 watts with parallel 801a’s. Havent checked the load lines for gm70 but should be similar. Think el84 should add some A2 power before A2 distortion becomes a big problem. I thnk weve proved on the 833a a small amount of A2 sounds rather good.

I can dust off the Tribute’s. Choke load the el84 cf dc driver. DC coupled means fixed bias obvious to some confusing to others.

[Paul Barker]

Tributes are already wired for 7k if memory is accurate looks like I’ll get 22 watts with +20v A2. More than enough.

Just on a rough estimate of 1,150v from minimum ac voltage to maximum ac voltage along the 7k loadline.

[simon]

Do you need 12W? The GM70 amp you heard at mine a few years ago had about 6W IIRC.

HT was just under 500V and had a simple LC coupled 12GN7 as triode driver with a gain of around 40. Dead simple, you'd build it in half an hour including finding the bits. I did have separate power supplies for the 12GN7 and GM70 as you're proposing.

I wonder if what you need right now is something simple, robust, but good sounding?

[Nick]

Cant help but wonder that the effort to make a new amp will likely be less that fixing the existing one.

[pre65]

My GK71 amp was heavily based (almost identical) to the Nick Chua GM70 circuit that sounds very like the one Simon mentions.

http://www.diyparadise.com/nikcgm70.html

[Paul Barker]

Cant help but wonder that the effort to make a new amp will likely be less that fixing the existing one.

That is true but that amp requires the el84 driver also to reach its potential.

Cramming the extra stage in the chassis means a new chassis and a new build.

So its not as big an advantage to starting again in all but the B+

Better just breadboard it though so it can be relatively quick and simple to build.

The +200/-200 can come from 50v telephone supplies stacked I already have and I can level the bias in 50v increments and the dc resistance of the choke leveling and flexibility in amount of A2 and the cathode current of the gm70 I can probably get a simple build.


801a’s are no longer made and ive got about 20 gm70’s four of which are copper, aswell as a load of 212’s that I dont plan to dust off at the moment.
Could make the VA really simple with a ecc83 6sl7 or mh41 as it only has to drive the grid of a cf, all the power drive is the job of the el84

[Paul Barker]

The 4k5 opts on existing amp can stay put and alter the amp to 45se into 4k5. New b+ suitable for 45 and its driver.

Could make that first. Stay sweet low power low voltage drive 45 with 10y with lower mu voltage amp maybe Marks favourite dhp.

3 direct heated stages start ac find out db of 100hz adjust it with s bench hum cancel on one stage only I hope.

The gm70 Ive already got covered with 20 volt 3.25 amp smps per channel hope it doesnt self destruct, may need chip cooling fan to make sure.

[simon]

You've a lot more nous and experience than me Paul but I couldn't get 3 stage AC filaments quiet, not even the 10Y driver. I chickened out and used good fil regs to remove my headaches.

[andrew Ivimey]

My gm70s use 20 volt 6 amps at least switched mode lap top psus from Maplin. Still going strong... I feel a bit of a cheat because they just ( still) work.

Dreadful as the phrase is, it is a 'no brainer'.

[Paul Barker]

Well the only caveat is the three stages musnt lead or lag no Lc coupling just RCR or ccs C R Then the sum of all three can be cancelled by one sbench hum cancel. Which in any case is adjustable. But you need to know which stage dominates and finish the cancelatiion of that at that stage. Might be the 10y. Critical not to alter the timing. Or youd have to cancel each stage individually. S bench cancelatiin circuits are much simpler to build than sexy dc circuits.and Ive never bought into dc heating is the only way. Ive never understood dc heating push pull amps.kiss.

[andrew Ivimey]

The point is... Gm70s want 3amps. If you are to provide 18-22 volts at 3 amps my experience is that you need to provide at least 6 amps or the circuit gets hot and conks out. Computer psus offer huge current range of useful voltages particularly if you enjoy Soviet transmitter valves and I don't seem to need chokes and other expensive devices - they just work. - exactly the same with 833s ...

[Paul Barker]

OK Andrew, I’ll have to look for 6a supplies then. Edit, Ive bagged two lenovo 20v 6.5A supplies for £15 each. That starts the project.

[Nick]

Then the sum of all three can be cancelled by one sbench hum cancel.

The fundamental maybe (assuming the mains AC is distortion free, which its isn't), but you will have a lot more of a problem canceling the distortion products of the hum signals.

BTW, I don't know why you think there is no phase shift in a RC stage. As long as you get the -3dB point down enough you may get away with it,

[Paul Barker]

Then the sum of all three can be cancelled by one sbench hum cancel.

The fundamental maybe (assuming the mains AC is distortion free, which its isn't), but you will have a lot more of a problem canceling the distortion products of the hum signals.

BTW, I don't know why you think there is no phase shift in a RC stage. As long as you get the -3dB point down enough you may get away with it,

Thank you for you’re insight. This applies to if I make the existing signal section into a 45 amp with all direct heating, which would be suck it and see scenario for experimenting evaluating and adapting. Im going to move away from that idea for now and look at the gm70 closer.

[Paul Barker]

I drew slightly better curves and came up with more acurate power output. Im thinking 650v B+ quiescent current 80mA 52 watt dissipation, even the copper GM70 OK with that. If voltage is the limit power out is greater holding current lower and existing power supply has operated well at that current drawer.

With those reasons for the operating position factored in, the minimum practical power out from 7k load is 17.85 watt 1000000/56000. requiring a 25v positive grid drive. The extreme is 25.71 watt 1440000/56000 requiring so +45 volts grid drive. I dont expect that peak to be atainable but somewhere between 17.75 watts and 25 watts Id hope for good sounds. Say that was 20 watts, it should suffice.

Yes of course I have much bigger power transformers, but if I use this existing power supply its maximum lump of iron I want in my living room these days. If I forklift my bigger transformers into an industrial facility I way aswell leave the 833a output stage already built and heard at Owston. Im trying to downsize. Back to Andrew I the heater supply already built into the 833a output stage hsn’t failed yet. To use it I have to shunt reg with a ty4/400 to make the smps start at all, its over powered for the task. Im OK with the ty4/400 doing that roll and a little final shunt regulation is no trial.