• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

Tubelab Universal Driver Board, 2015 version

Sorry for the vanishing act, I have been out of town with only phone access to the internet for two weeks. I stopped by ESRC for a large box of tube sockets, but for the first time in 20 years I went there and came away without buying a single tube!

It seems to me that 350V on the V2 anodes would be more appropriate.

It depends on what output tubes you are running and how much drive they need. I simply tuned my set up for minimum high order distortion and measured the voltages. I would assume a different setup with different tubes may achieve lowest distortion at a different voltage.

You are probably right that the voltage should be lower, particularly with a 450 volt or lower supply voltage, but driver tubes have 150 volts or so on their cathodes and need some voltage across them.

I am in the process of putting my test system back together so I can take some measurements with different output tubes and supply voltages.
 
Puzzling THD behaviour

Hi. Good to have you back!

It turns out that I've been able to duplicate in my sim exactly what you measured and I'm not completely sure why.

My LTSpice schematic is shown below with my current preferred cathode feedback output stage.

I set up the operating points for the LTPs by minimizing the THDs they generated on the assumption that the lower the THD arriving at the output stage the better.

I started wondering what would happen if I raised the operating point of the second LTP towards the voltage you'd recommended.

Sure enough, the THD from the second LTP was much higher (I put the 2 operating points on the 6CG7 plate curves and it's easy to see why). Gain was also reduced and again, easy to see why.

But what was really weird was that as the distortion produced by the second LTP increased, THD at the output decreased!!

Part of the reason has to be that the output stage cancels odd order harmonics but there has to be more to it.

I wondered if there was a sweet spot and plotted V2 Anode THD and Load THD vs V2 Anode Voltage (second screenshot for an input that gave me 40W out for a V2 Anode voltage of 342V).

Sure enough there was! At a V2 anode voltage of 420V.

As I said gain decreases as you increase the V2 Anode voltages and consequently the power out decreases. This is shown in the third image.

The final image shows the THD vs Power out for V2 Anode voltages of 340V and 420V. The 420V operating point provides a far lower THD up to 50W or so but is limited to 60W if you want < 1% THD. If you're prepared to put up with higher distortion there's good power out to 90W with the lower V2 Anode operating points.
 

Attachments

  • LTSpice schematic.png
    LTSpice schematic.png
    217.5 KB · Views: 628
  • THD vs V2 anode voltages.png
    THD vs V2 anode voltages.png
    69.3 KB · Views: 625
  • Power vs V2 Anode Voltages.png
    Power vs V2 Anode Voltages.png
    38.3 KB · Views: 609
  • THD vs Power.png
    THD vs Power.png
    61.7 KB · Views: 607
Part of the reason lies in the fact that this is a class AB push pull amp. Each output tube contributes an equal share to the total output averaged over time, but not at the same time. The tube that is entering conduction begins to contribute some to the total output, as both tubes are conducting equally they contribute equally to the total output, and as that output tube reaches peak conduction it is contributing 100% of the output power because the other tube is cutoff.

Note that this is less true on a class A push pull amp since both output tubes are always conduction, but the share of total work is constantly changing.

The driver's contribution to total distortion is also not equally split between the two tubes. The tube that is driving a particular output tube into conduction contributes the most to the total amplifier power output and distortion at that moment.

As the driver's plates go lower in voltage the corresponding output tube is contributing less and less to the amps total power output and therefore distortion. I have seen drive signals with severe clipping on the bottom due to the mosfets hitting the negative rail but the amp's distortion doesn't change much.

I avoid clipped or severely distorted bottom halves due to limited negative rail voltage for another reason. The mosfets tend to emit a burst of high frequency energy when is goes into cutoff or comes out of cutoff, so I like to keep ample headroom across them. You can see this in a scope, hear it, and severe cases fry a tweeter with ultrasonic energy.

The THD of the driver stage can be higher than the amps total THD even without feedback. This baffled me for several years since I would always test and optimize the driver circuit before building the output stage, then wonder why the driver worked better with a different tune up once the output tubes were added. The top half of the driver's output is far more critical than the bottom half. This makes optimizing the driver board alone with distortion measurements rather pointless.
 
Thanks for the explanation. It certainly is counter-intuitive that increasing earlier stage THD should decrease output THD.

What you're saying is that the positive half of the waveform is amplified with less distortion than the negative half and as you raise the operating point of the second LTP the distortion introduced to the positive half reduces.

Pity about the associated loss of gain.

I'll play around with global negative feedback for the 341V and 420V 2nd LTP anode operating points.
 
Last edited:
It certainly is counter-intuitive that increasing earlier stage THD should decrease output THD.

As I said it confused me for a couple of years.

I was also quite baffled by the effects of feedback, especially feedback from the output tube plates to the driver or pre driver plates. You can connect distortion measurement equipment up to the driver board output and connect a second set of measurement equipment up to the speaker terminals. Then apply some sort of feedback to the amp that reduces it's total distortion. The distortion at the output of driver stage almost always increases.

I didn't understand this and spent considerable time trying to fix it until my day job of designing RF power amplifiers took me down a path called adaptive pre-distortion. Here we purposely add distortion to the baseband LTE signal in the DSP such that it "bends" the signal in a manner opposite to the "bending" occurring in the RF power amp.

I use the term "bending" to describe the nonlinear transfer function of a "linear" RF power amp. This distortion in an RF amp is not constant and varies with the power level, modulation, temperature, and load impedance. This bending occurs in both amplitude and phase.

We can guess, and apply some correction based on testing a bunch of power amps and gain some improvement, or we can build a closed loop Cartesian Feedback system and change the pre-distortion algorithm on the fly for considerably more improvement. Both techniques are currently used in cell phone and cell tower systems.

We can see a similar effect in a two stage SE amp using triodes for both stages. A simple SE triode amp will have a bent transfer curve, IE it's gain at strong signal will be slightly less than with a mid level signal. This results in mostly second harmonic distortion. Connecting two such stages sequentially can have a lower overall distortion than each individual stage. A 6SN7 driving a 300B can exhibit this effect, that's why that combination was so popular years ago.
 
Hi George,

Welcome back. I have a couple of questions related to my bench test and the startup procedure. (See post 249) I'm getting ready to bench test again and
I am wondering if I can adjust the bias etc using only cheap multi-meters or do I need more sophisticated equipment?

thanks, Jacques
 
Good to see you back George.

I finally got around to building the dual tracking power supply for biasing the source followers, so I should be able to get back to testing before too long.

It looks like it is good for 60mA and I can adjust the negetave supply down to about -95V. Tracking isn't great, but it shouldn't be critical since the actual bias adjustment is separate. Hopefully it will be good enough for the universal driver board. I can reduce the R in the CRC to increase the output under load, but would need to change the 100V reference Zeners to a greater value to get more compliance.

I still need to box it up.

Summer is slow for electronics. I'm still trying to get my Sportster back on the road.

I have camp outs planned for two of the next five weekends, and there is the Shelby Ham Fest Aug 31-Sept 2.
 

Attachments

  • DSCN3128.jpg
    DSCN3128.jpg
    1,001.8 KB · Views: 480
Last edited:
Jacques,

You should be able to get the amp working with cheap meters unless there is a problem. You say that "the sound cuts out if I raise the bias above 15 mA." Does this happen abruptly, and does the current change abruptly at this point? If so the amp may be oscillating, which may require a scope to find. Try cleaning up the wiring to see if the point at which this happens changes.

If the amp begins to distort and then cuts off as 15 mA is approached, then I would try turning the voltage dowwn via R13 to see if a happy point can be found. How much B+ do you have, and does it drop significantly as the bias is turned up?
 
output tube diagram.jpg
Hi George,

We were away riding our bicycles thru Ohio while raising money for trees. We rode 560 miles with 26,000 feet of climbing over 6 days -- I didn’t think Ohio would be so hilly. We are back, but I’m still recovering. Anyway I’m back to testing my UDB. I believe I have a problem with the way I’ve wired the output tubes. My B+ is just at 500 volts before I add the output tubes. Once the output tubes are in it drops down dramatically to 250 volts. Can you look at the drawing attached and let me know if it looks OK to you? I’m getting sound if I keep the voltage at R13 set to under 250 volts and the bias around 35ma.

Thanks, Jacques
 
George, if I want to run GNFB from the output of the OPT do I go to pin 7 of V1B and swap DRV2=2 and DRV 1-1 as necessary to get phasing correct?


Any recommended change to C1 if I do? Looks like it may affect LF response.

I'm about ready to start torturing some tubes. 6P3S is first up, to be followed by 6P41S.
 

Attachments

  • DSCN3144.jpg
    DSCN3144.jpg
    789.2 KB · Views: 335
Last edited:
George, if I want to run GNFB from the output of the OPT do I go to pin 7 of V1B and swap DRV2=2 and DRV 1-1 as necessary to get phasing correct?

For global negative feedback George recommended adding a 100Ω resistor between the 0.47uF capacitor and ground and running the negative feedback via a 5k resistor to the junction of the capacitor and resistor.

In my simulations I've run a 2.5K feedback resistor without a problem. Gave me about 10dB of feedback I think.
 
Last edited:
After finally finding the time to focus, I found my mistakes and now I have one board working just fine. I’m running El34’s with the bias at 30ma. B+ is at 460.

I have a few questions. Per George’s set up instructions I set the plates for V1 at 110 volts and V2 at 450 volts before adding the power tubes. With the power tubes in place, the two plates in V2 drop to 400 volts. Should they be adjusted up?

Without a scope I guess the best way to dial in the bias and voltage adjustments is just to listen.

That being said, I’m thinking of a purchasing an oscilloscope. Anyone have any recommendations? New, used? Is an old Tektronix the way to go? Will a new compact one do the trick.

Thanks, Jacques
 
I'm a bit reluctant to offer troubleshooting advice on George's board (actually I'm not sure I should be providing advice to anybody!) but with a B+ of 460V (which I assume is what you have on the OPT centre tap and the driver anode resistors), 450V seems too high for the driver anodes being only 10V below the supply.

Check the B+ voltage after installing the output tubes. Perhaps it's dropping more than it should and causing the driver anodes to drop.

Without a distortion analyzer I'd suggest setting the V1 and V2 anode voltages as George has suggested and set the EL34 biases with a meter to the recommended value and leave it at that. A mA or three up or down probably won't make any audible difference.

As for a scope, I bought a cheap Chinese one (a Rigol). It meets my hobbyist requirements. For messing with high voltages make sure you get one or two high voltage probes as well.
 
Initial setup has been going well. Although somehow I smoked the 10K resistor at R23. Voltage drop is 110V for a power dissipation of 1.3W in a 2W resistor, so I'm befuddled.

Plate of the input 6N1P-EV came in around 110V, where I want them.

6CG7 plates are around 210V at 11.7mA, cathodes are at 101V for a Vak of 109V.

I have a funeral to attend for the Aunt of my ex tomorrow, so I'm breaking off for the night. The aunt was a nice old gal, lived to the age of 95. Not a bad run, we all should be so lucky.
 

Attachments

  • DSCN3145.jpg
    DSCN3145.jpg
    991.7 KB · Views: 305
Last edited:
I'm a bit reluctant to offer troubleshooting advice on George's board (actually I'm not sure I should be providing advice to anybody!) but with a B+ of 460V (which I assume is what you have on the OPT centre tap and the driver anode resistors), 450V seems too high for the driver anodes being only 10V below the supply.

Check the B+ voltage after installing the output tubes. Perhaps it's dropping more than it should and causing the driver anodes to drop.

Without a distortion analyzer I'd suggest setting the V1 and V2 anode voltages as George has suggested and set the EL34 biases with a meter to the recommended value and leave it at that. A mA or three up or down probably won't make any audible difference.

As for a scope, I bought a cheap Chinese one (a Rigol). It meets my hobbyist requirements. For messing with high voltages make sure you get one or two high voltage probes as well.

Thanks for you advice. I found that everything is falling into line on both of my boards, so I'm good to go. I had some confusion about how to measure the bias, but that is now cleared up and things look good. I decided to ignore the V2 plate voltages once the power tubes were in place.

Which Rigol oscilloscope did you purchase? For working with tube amps will 50mhz suffice? Are two channels enough? Are there any other parameters I should be concerned with?

thanks, Jacques
 
V2 plate voltage should be about 1/2 of the difference between the V1 plate voltage and the supply voltage. This gives you maximum swing between cutoff and saturation. That said, minimum distortion may be found somewhat off of the 1/2 value. The power tubes will affect the supply voltage as they can draw more current than the pre and driver tubes.

In my case, I'm using an HP 6844B power supply which is not effected by loading at the level I'm testing at. I will have to be careful with my power supply design to insure that I ultimately operate at the same bias points as my bench tests. I will be doing several months (spread out testing) of different tubes to see how they play. Right now I'm burning in and testing 6N3C tubes.

The only advantage of having a scope with a wider BW than 50MHz is you can see higher frequency oscillations, which occur rarely. I use a 100MHz B&K 2190E because it was in my price range. Most of the time I set BW limiting to 20MHz just to eliminate RFI because I'm close to tv Transmitter towers. Two channels are generally enough. I haven't used my TEK 2465 in years. just don't need it. FFTs are better done on a PC.