• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

#26 pre amp

Have replaced the cascoded MOSFET gyrator with Anatoly's hybrid gyrator which performs better at lower anode current (bearing in mind a 26 or 01a stage)
It looks a lot of silicon to me but should work really well.
Some questions:
1) Probably the gyrator led array should be increased to allow at least 10-15V across M1 and Q1? In a previous version for my 45SE driver I used zeners instead, but would like to stick to LEDs where possible.
2) is the Mu follower configuration ok with this gyrator design? Wasn't sure as Anatoly didn't include it in the original circuit diagram. I added the resistor between the anode and the gyrator with a value of 1/gm

Also added additional LEDs to the follower's CCS. I think a quiescent current of 10-15mA should do fine to drive the amp's capacitance?

Will
An externally hosted image should be here but it was not working when we last tested it.
simulate this in SPICE when I return from my trip...
Thanks
Ale
 
Ale, this looks like a good circuit.

10mA should be fine for the cable driver, for reasonable cables and input capacitance of the power amp. If in doubt, use the capacitance value of the cable you have in mind (100pF = normal 1metre, 1000pF = excessive, like flat cables) and calculate the cable current at 50kHz:

For 100pF cable + 100pF amp input, Xc is 1/2ПfC = ~16Kohm
Say peak output voltage is 10V, peak current is 600uA for 100pF cables,
For 1000pF cables and 100pF amp input, the current peak is more like 3,5mA !

The current source should probably sink more than 10x the peak current you have in mind, so I think it's usually worth checking the cables you have.
 
Hi Rod,
Thanks for the input! My amp has probably about 100pF input capacitance (6j5 input driver at Mu=20 and Cag=4pf) and use 2m of normal cable so another 200pF.
Why considering current at 50kHz and not 22KHz?

22kHz can be substituted here, but I use 50kHz -
- to ensure a robust design margin;
- if an LP player using Moving-Coil cartridges is ever connected, this will be the bandwidth! Especially, tracking over big clicks will generate this kind of spectrum, and we want the preamp to remain in control!

Why also driver's sinking capability needs to be around 10x if we are already considering the peak current?

The Source-follower can only source current (ie charge the cable capacitance with current flowing OUT of the amp. When the signal voltage heads downwards, the CCS must deal with the current. Since the CCS current is fixed, for every mA it takes from the cable, the source follower decreases in current by 1mA....

If we set the follower's idle current equal to the PEAK output current, then the follower current will vary from 0.1 to 3mA (say), or 1mA to 4mA if we leave a little standing current.
Sweeping the follower current over a 4:1 ratio will make big changes to the FET's output impedance as the signal progresses, and also the gate-source voltage (introducing serious distortion).

This is the price to be paid when doing away with feedback in an amplifier. The design must be such that the signal can sweep across its maximum extent without changing the operating conditions of any of your stages too much. This is a big problem for solid-state design, because transistor are horribly nonlinear.

Of course, in the commercial world, feedback is used to 'linearise' the amplifier by brute force.

If you can get a 10:1 ratio of standing-current to peak current (at 50kHz), the performance should be OK. If 30mA is achievable (within the safe-operating area of the transistors!) then this is even better. But beware than SOA and power handling need to be double checked when voltages of 20V or more are used - it is very easy to overstep this in an amplifier.
 
I'm still waiting on the SSHV2 boards before I start this DHT pre/headamp project. But Rod has just elequantly described why the #26 needs an output tube for a pre with gain. Frank Cooter was one of the first to delve into this and for an all DHT headamp/pre he had the first DHT loftin-white followed by a 71A 5k:600 output. I think he was using 112A-71A at the time. But Moglia's lower filament current 4P1L with a mu near 10 with a 5k primary OPT would just about leave enough gain to keep it spud.

The #26 as a spud even with an OPT and just as a pre (forget hp-out) seems troublesome because of the high Rp and low current.

The big question for me has been does a #26-71A sound better than a 4P1L spud. Potentially the #26-71A would have some distortion cancellation as loftin-white, but there is always something to be said for using a single tube. And the cost of adding the 71A is high. The personal issue with the 4P1l is it isn't a "classic" triode and that extra grid there that is in the way seems like it would be a negative to SQ.

But the SSHV2 boards will be here soon, time to make a decision.
 
HI regal I don't know if is the same but on 46 the extra grid is no problem for SQ...just different then 45

I tried out the 31, 71A and 46 as stages after the 26. Didn't like the 12A, incidentally. The 31 is pretty good, the 71A a little better, but with the 46 you get a better sound than both and more gain. So the 46 was my choice. I have hopes for the 4P1L - if that sounds good the gain and low anode impedence will be very useful.

andy

Thanks for the feedback, right now I'm using a WE417A SET OPT, its a bit strident for lack of a better word, looking foward to trying the DHT's.
 
Thanks Rod for the explanation. I will check when I get back home which MOSFET I have at hand and their Pd capability. If I can run 30mA will do.

The 4P1L with a gyrator load is ideal as has low anode resistance (1-2K in triode mode) and also is biased at 20mA which makes it a good candidate to drive the cables and the amp's input capacitance. Also 71a performs really well.

4P1L is a challenge due to microphonics. Providing the valve is suspended properly and also filament are starved accordingly is a great valve for a preamp.

I liked the sound of 01a and other DHTs with greater anode resistance and low quiescent current. Therefore I want to test whether with the source follower (or a cathode follower if anyone fancies the addition of an additional valve including the filament supply required) con overcome the challenge and provide a better response of these valves....

Now I got to the point that with same power supply can do a side by side tests between different DHTs which is great to listen to differences and optimise the setup.
Cheers,
Ale
 
Ale,

If you want to try a cathode follower, please don't use a DHT for this. That's not to say I don't like them (I surely do), but in any follower you want low output impedance. This translates into HIGH gm.

I would recommend an indirectly heated valve in this position.

What about this idea -

26 DHT in filament bias - but instead of a filament bias resistor, use the heater of 3x Mazda 30F5 as part of the bias circuit! The 30F5 needs 7.3V 300mA to heat, so you could run three of them in parallel, and add a little resistor to make up another 2V (approx). Yes, use some heaters as a filament bias resistor, the economy of it is DIY mighty!

This way, you get filament bias without needing a heatsink, and the cathode follower is heated for free.

If you triode connect the 30F5 you get nearly 10mA/V each, which with three paralleled - equates to a output impedance approaching 30 ohms!

Run these into a power-FET CCS with 40mA of standing current finishes the output stage.

If you can't get 30F5s, there's lots of cheap Russian triodes & pentodes (with high gm) that will do equally well.
 
Hi Rod,

I like your idea very much. Do you think it's OK if instead of using 3 x 30F5 which is difficult to obtain, we could use two 8CG7 (cheap on ebay) with heater in parallel (heater is 8.4V and 450ma) and a little resistor to make up for another 0.6V (or no resistor at all) and use one of the 8CG7 as cathode follower for each #26.
 
Hi Rod,
That's a brilliant idea. I remember now a similar circuit in a tiny headphone amplifier where the valve filament was fed from the output source follower.

Any particular reason why you chose the Mazda valve? Found a set of valves in eBay which I'm bidding for but they are not easy to get hold of. May need to look for some Russian pentodes instead!

I think this is a circuit variance which is very easy to implement, can try and listen to the results
I may adapt this one for my 01a preamp which I have currently breadboarded and will probably need only one valve per channel, so is dead easy to implement.

Would you run a cascoded MOSFET CCS (eg dn2540) or a simple power MOSFET (eg IRF840) instead?

Thanks
Ale
 
Hi Rod,

I like your idea very much. Do you think it's OK if instead of using 3 x 30F5 which is difficult to obtain, we could use two 8CG7 (cheap on ebay) with heater in parallel (heater is 8.4V and 450ma) and a little resistor to make up for another 0.6V (or no resistor at all) and use one of the 8CG7 as cathode follower for each #26.

The 8CG7 could be used, but it is a little short of optimum. The gm is only 2.6mA/V on these, and as a 6SN7 type structure, they prefer 300V rather than the 150V we probably have in a 26 line stage.

Russian pentodes, triode-connected are a good bet. The 6Ж5П (6J5P) is rated for 150V operation and 9mA/V in pentode, which should mean over 10mA/V triode, and the heater is 6.3V 300mA.

you could also do the job with a single bottle:
The Russian 6H6П (6N6P) is a twin triode with 6.3V 900mA heater, and slope of 11mA/V.
Ebay seller boreca-75 and others have these cheaply. These should give excellent performance.

Please remember to use a grid-stopper though, starting at 10K, maybe going even higher.
 
Hi Rod,
That's a brilliant idea. I remember now a similar circuit in a tiny headphone amplifier where the valve filament was fed from the output source follower.

Any particular reason why you chose the Mazda valve? Found a set of valves in eBay which I'm bidding for but they are not easy to get hold of. May need to look for some Russian pentodes instead!

I think this is a circuit variance which is very easy to implement, can try and listen to the results
I may adapt this one for my 01a preamp which I have currently breadboarded and will probably need only one valve per channel, so is dead easy to implement.

Would you run a cascoded MOSFET CCS (eg dn2540) or a simple power MOSFET (eg IRF840) instead?

Thanks
Ale

Hi Ale, The Russian 6H6П (6N6P) twin triode I mentioned above, is probably a better solution, and only one bottle is needed - now that I have taken a second look. Lots of valves will work in this position, but they will mostly be 1950s designed TV valves, intended for 150 to 180V operation, and with gm above 7mA/V

For the CCS you'll need 40-45mA for typical pentode (x3) or 60mA for both halves of a 6H6П. The cathode followers will operate with 130V to 160V (according to chosen type) across them. If we allow 30 to 50V across the CCS, then the dissipated power in the transistors will total 1.5W, which is no problem. I would start with the NPN bipolar CCS (with LEDs) that you drew earlier, and use a nice 250V NPN in a power package - TIP47, MJE340 etc. A tiny heatsink will suffice.
 
Hi Rod,
Quick update to my 01a preamp attached. I know the MOSFET gyrator is not optimal for a low anode current, should change it to Anatoly's when I get the chance as per my previous diagram. Anyway, this is a quick setup I can test easily.

Looked at SMD bipolars as was thinking to get a very small PCB board including SMD LEDs here.

I hope 15mA of idle current at the CCS should work fine in this case?

Thanks
Ale
 

Attachments

  • 01a preamp01.png
    01a preamp01.png
    112.1 KB · Views: 1,344
26DHT stage with source follower

Rod,
Looking for some help here. Attached is the simulation of the 26 DHT with mosfet source follower as discussed. Frequency response is better than a 26 with a 300H choke. However, distortion is slightly worse 0.07% against 0.05%. This is due to the gyrator which I can't set properly due to the HT level I guess. I think if I were able to increase the HT above 200V should be able to provide headroom to the PNP to enable bias at 170V and 6mA.

Cheers,
Ale
 

Attachments

  • 26dhtgyrator-follower.PNG
    26dhtgyrator-follower.PNG
    41 KB · Views: 1,312
  • 26DHTgyrator-follower-freq.PNG
    26DHTgyrator-follower-freq.PNG
    42.6 KB · Views: 1,244
Andy,
Here is a quick sample of the impact of filament starvation on the DHT anode resistance. I have traced the 26 DHT with 1A and 700mA. You can see that the anode resistance increased from 8K to about 11K

A higher anode resistance is a clear drawback if you want to starve filaments with a choke / gyrator load to drive the cables and amp's input capacitance directly :(

cheers,
Ale
 

Attachments

  • 26 Fil=1A.JPG
    26 Fil=1A.JPG
    46.6 KB · Views: 1,047
  • 26 Fil=700mA.JPG
    26 Fil=700mA.JPG
    46.3 KB · Views: 974