PP question - best driver stage!

[andyjevans]

I'm currently building a PP 300b amp with O-netics OPTs. I have a chance to build something really good here. It will be all-DHT so that's a final design decision - non-negotiable I'm afraid! I have loads of DHT driver tubes of all types - 71A, 10Y, 31, 46, 1624, etc and also 26 which might be usable. There's little I don't have in DHTs, even the European ones like RE134.

I also have several Lundahl interstages, like LL1660/PP, LL1671PP, plus LL1660/5mA, LL1660/20mA which could be used as phase splitters. I also have loads of Russian teflon coupling caps.

I've had great sound with a SE input stage with IXYS 10M45 cascode active load and red LED in the cathode. Love it and want to keep it. Currently 1G4GT but could be 26 or 01A.

So, cooks, there are your ingredients. I have the chassis built ready. Which of the above would you use? Your mission should you choose to accept it is to negotiate your way from a SE input to 300b outputs! I have about 320v available for the driver stage at up to 200mA.

andy

 

[Bandersnatch]

hey-Hey!!!,
This is not enough voltage for what I have in mind. Probably better suited to a pair of 845... In any case, look at type HY40. I figured a B+ around 750V would do well( for resistive load ). Now if you got a 1+1:1+1 IT, the '40 might be able to run from that low voltage. Consider raising B+, perhaps use the Blumlein Garter to bias/balance them( and add ~150V to B+).
cheers,
Douglas

 

[Shoog]

Personnally I would go with the LL1660/PP and drive it with a LTP at the front. This should give you the maximum drive potential with no chance of blocking distortion, and no caps in the signal path. If you need lots of gain then consider one of the other Lundahls up front to drive both branches of the LTP and maintaining all the gain of the driver rather than halving it.

Shoog

 

[Shoog]

hey-Hey!!!,
This is not enough voltage for what I have in mind. Probably better suited to a pair of 845... In any case, look at type HY40. I figured a B+ around 750V would do well( for resistive load ). Now if you got a 1+1:1+1 IT, the '40 might be able to run from that low voltage. Consider raising B+, perhaps use the Blumlein Garter to bias/balance them( and add ~150V to B+).
cheers,
Douglas

The Garter arrangement works extremely well for balancing current and should be considered for the output stage. There is a version with transistors which gets over the issue of burning up twice the voltage in bias (a big issue with 300B's).
In a LTP driver stage the Garter would undermine the LTP behaviour and could only be considered if both input of the LTP were driven. Consider putting a CCS below the garter and referencing the bypass caps to its anode, this would create DC current balance whilst maintaining LTP behaviour.

Shoog

 

[andyjevans]

Hi guys1

I'm ruling out high voltages - 450v or so for the 300b is quite enough. I could use that voltage for the driver, but I have a separate transformer for that giving about 320v - less if I use choke input which is an option.

So, yes, LL1660/PP is an option - could even use 10Y as the LTP or 71A, 31 etc.

If I cap couple from the input stage, I ground the second grid of the LTP, yes? that's what I did in my current amp.

If I direct couple can I do the same or should I use a 1M resistor from first grid to second grid and a cap to ground like the Mullard circuit?

andy

 

[Shoog]

Does the maths work out to make this a two stage design if all of the gain of the driver is available to you?

Shoog

 

[andyjevans]

Hi Shoog,

I really need three stages with DHTs. Nothing over mu=15 here. Plus driver will be anything from 3 to 10, closer to 3 probably. In any case I'll probably need a preamp, but that's OK - have a nice 26 preamp. So gain isn't critical, but it does need to be more than 2 stages with DHTs.

andy

 

[andyjevans]

I should add I'm driving Apogee ribbons - Caliper Sigs. In a smallish room at low volumes so I already know PP 300b works. But of course, sensitivity is around 84db, so do need 4 stages in all. Right now I have 1G4GT (mu=15) into 6SN7 into 807 and that's enough gain for three stages, but I won't get a driver stage with as much gain as 6SN7s and as said, this is going to be strictly DHT!

andy

 

[Shoog]

have you looked into the Karna. Seems as if you have the iron to pull it off.

http://www.nutshellhifi.com/triode2.html

Shoog

 

[andyjevans]

Addendum - i might possibly, possibly (don't know and don't really want to) use a pair of 2C22 (6J5 type with 2 top-caps) for the LTP. That would give me the same gain as a 6SN7. It's about the only indirectly heated valve I could imagine using. That could work without a preamp. But more likely 3 stages of DHTs

andy

 

[andyjevans]

I do indeed have the iron for a Karna type all transformer design with 3 stages. I'd quite like to only use one transformer, though. Those Lundahls are expensive and I'd prefer to use them in future projects! Plus if I used three PP stages it would be really heavy on filament supplies for the DHTs - 6 DC supplies! Arrgh! It can be done, but I'd like to simplify a bit.

andy

 

[JLH]

I wound go 46 triode wired with the LL1660/20ma configured for SE to PP. Then use push pull 10Ys with LL1660/PP into the 300Bs. You could spare the LL1660/20mA by CCS loading the 46 into a parafeed autoformer phase splitter - should be less money than a LL1660. I've always through oxide filament tubes sound better driven with tungsten filament tubes, and vice versa. My two favorites are the 46 driving a 845, and the 10Y driving a 300B. Just my $0.02 worth.

Rgs, JLH

 

[andyjevans]

I really like the idea of 10Y through LL1660 into 300bs. I've tried that combination, and I think there's nothing finer. I think we're singing from the same hymn sheet here!

For the input I'm more inclined towards a higher mu valve less hungry on filamants - current favourites are 26, 01A, 30sp (VT-67) and 1G4GT. Active load and parafeed sounds perfect.

Suggestions for a parafeed transformer? Or really only needs to be an autoformer. I have some Hammond 124b but they're 90k to 10k stepdown, and want 1:1.

I could cap couple of course, and ground the second grid on the 10Ys. Might do that while waiting for a good autoformer.

andy

 

[zigzagflux]

The triode wired 46 - LL1692A - 300B has worked very well for me; I am able to drive the 300B grids slightly positive without distortion. This at 300V B+ and 28mA per triode. Consider Kevin Carter's design at K&K if you want to save that additional IT. DC coupling from either a choke or an IXYS cascode CCS in the plates of your 26 or 01A should do well. You could also put a CCS in the tail of the 46's.

Are you sold on DC filaments? With shorted input, all AC heated, I run less than 0.5mV rms of hum on 4 ohm speakers.

 

[Shoog]

Autotransformer phase splitting should work really well, but since you are going for a no holds barred ultraproject it seems a real shame to start throwing in coupling caps. If this works I think you will probably never have to build another amp again so for me the extra iron seems worth it.
I think SE-Interstage-LTP-interstage-300B-PP would be a fine compromise on the Amnity.

Shoog

 

[Wavebourn]

Here is mine; add 240K feedback resistors from plates of output toobs to plates of driver toobs (between 360 Ohm and 17K resistors), then add a GNFB from speaker to the 1'st tube's cathode. You may use GU-50 or KT-88 outputs with regulating screen supply, the result will be as the triodest as possible output triodes, though connected as pentodes.

http://wavebourn.com/forum/download.php?id=119&f=7

It is the cleanest driver I ever designed, to load on low input impedance of parallel feedback by voltage.
Servo from tails of a driver LTP stabilizes the whole thingy applying DC feedback to screen grids of the input stage tube.

Enjoy, it's public domain, even though I am going to sell Pyramid amps!

 

[andyjevans]

Many points to ponder here. Wavebourne's solution is beyond my knowledge so I'm probably not equipped to go there, but very interesting! Shoog's SE-interstage-LTP-interstage-300B is certainly good. But I wonder why I'm hearing better results with a IXYS active load than a transformer? Maybe my memory is playing tricks, but I just love the sound of the active load. I'll certainly have to build some different versions of this, and report back to you guys!

AC filaments. could be done on the 300b. Would be noisy on the driver, though wouldn't it? Wouldn't a 46 hum with AC as a driver?

Lastly, with my PP 807 amp I tried the outputs with a plain resistor, then bypassed with a 47uF polypropylene to the HT on the centre tap of the OPT, then bypassed to ground. I think I liked the unbypassed resistor - sounded purer and cleaner. Does that make sense, guys?

Keep the ideas coming!! Hope to build some this week.

andy

 

[Wavebourn]

Here is already tested version of an active load, you may want to try it.

http://wavebourn.com/forum/download.php?id=125&f=7

 

[TonyTecson]

hi tolec,

is your topology amenable to higher voltages for greater output swing? say from 400 to 600 volts? what is the output swing with 400volt supply? thanks....

 

[Wavebourn]

hi tolec,

is your topology amenable to higher voltages for greater output swing? say from 400 to 600 volts? what is the output swing with 400volt supply? thanks....

Voltage and power is mostly limited by PNP transistor. Everything depends on which one you can find. However, you may use a single N-channel transistor (as in SRPP) instead of all of this, but dynamic resistance will be lower, and will depend on frequency more.

By the way, here is an improved version I designed for Hearinspace:


http://wavebourn.com/forum/download.php?id=125&f=7