• 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.

Impedance multiplier hybrid parafeed

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Here is a generalization of the anti-triode concept. It's a hybrid with a tube voltage amplifier followed by a true current amplifier. The current amplifier is not a simple voltage follower, but a power stage with a defined current amplification ratio. A defined fraction of the output current is handled by the tube stage and the rest is supplied by the output transistors. I've scaled this so it can use a PA matching transformer as a parafeed OPT.

This allows the actual speaker load line (or OPT primary load line) to be scaled up in impedance to the voltage amplifier current range. The tube sees a constant fraction of the output load, allowing the tube to respond to the speaker in classic SET fashion, but at a higher output power level. Thus the characteristic is imparted by the tube, with the output power scaled up by the current amplifier stage. Another way to think about it is an impedance multiplier which scales the output impedance up by some defined ratio so the plate of the tube sees a scaled version of the actual loudspeaker load.

I built a simple op amp controlled circuit to test the idea.

The first stage is a SE totem pole anti triode with the idle plate voltage arranged to be near the midpoint of the power supply voltage so as to allow for symmetric swing. The anti triode is a 2:1 current booster or a 2X impedance multiplier.

The output node of the anti triode is directly coupled to the output stage such that a 3 way current split is achieved between the tube, the anti triode MOSFET Q1, and the output stage. Current is split according to the ratio of resistors R6 to R4 & R5 in parallel, here about 10:1 resulting in a 5K bootstrapped load at the anti triode output node, which results in a 10K plate load on the triode, scaled up a total of 20X from the 500 ohm OPT primary load.

Another way to think about it is that the triode is responsible for the characteristic transfer function and plate resistance by driving a defined fraction of the load. The MOSFETS handle most of the current while letting the tube work into a scaled version of the speaker load.

The output impedance is approximately the plate resistance divided by 20 or about 50 ohms, giving a damping factor of about 10.

The op amps are CCS powered and track the output voltage. The op amp feedback is arranged to provide a precise signal current split at all times.

Bias is servo stabilized at 100 mA on each side separately using the 1.2V drop across a diode connected darlington transistor and the 12 ohm source resistors.

The output stage operates in class A for most of the signal swing. The full signal output voltage with a 325V power supply is 100V RMS, which allows this amplifier to drive a 100V PA matching transformer directly. I use UTC LS-33s because I have them but there are others on the market; some are of really nice quality and have full range response.

The prototype is scaled to 20W power output, which is 100V RMS into 500 ohms. Selecting the proper taps on a 20W matching transformer for the speaker impedance will result in approximately 500 ohms load. 60uF coupling cap gives an Fc of about 5 Hz. The circuit could be scaled to use 70V matching transformers as well, with lower B+ and higher idle current.

Next step is to do some measurement of the distortion spectrum at various power levels.
 

Attachments

  • ImpMultHybrid.png
    ImpMultHybrid.png
    20.9 KB · Views: 570
  • amptop.jpg
    amptop.jpg
    44.5 KB · Views: 553
  • ampboards.jpg
    ampboards.jpg
    93.9 KB · Views: 528
Last edited:
...Is this a PA amplifier ? :)

No.

It's a full range hifi amplifier with a power frequency response of 10 Hz to > 50 KHz and a SET harmonic structure with no distortion cancellation of any kind.

Using a matching transformer doesn't necessarily relegate it to PA use, if you take care to use a good full range transformer.

Also is just in the range to get 20W out of a 220/44V power toroid if you scale back to 70V output with a little higher idle current.
 
Last edited:
Looks elegant. What will happen if you instead of the antitriode output of the imput stage take the signal from the source above the two 91ohm resistors? I suspect distortion will be a lot lower that way with a lowish 5 kohm load.

Right. The Zout gets really low in that case as does the distortion. I can also accomplish the same thing by increasing the current gain of the output stage. I can dial in different amounts of SET-ness i.e. load line steepness.
 
Last edited:
Michael, nice circuit.
Does it behave well on power up? (R23,R24 controlling that I guess). Could this be biased for class AB? (each impedance converter working half time) Maybe could just use a CCS for Q5 (instead of the anti-T), since the rest of the circuit is already impedance multiplying.

It would be interesting to also adapt the circuit for a conventional P-P setup into a CT'd power toroid OT.
 
Last edited:
Michael, nice circuit.
Does it behave well on power up? (R23,R24 controlling that I guess). Could this be biased for class AB? (each impedance converter working half time) Maybe could just use a CCS for Q5 (instead of the anti-T), since the rest of the circuit is already impedance multiplying.

It would be interesting to also adapt the circuit for a conventional P-P setup into a CT'd power toroid OT.

It does power up OK, but I have been powering up the heater first with a separate supply.

I want to play with the SE harmonic signature at varying levels as well as validate the AT behavior wrt 2H, impedance, and clipping.

CCS and mu-follower configurations are easy to do by just changing resistors around. I will be interesting to compare AT and CCS with 2x the output stage current ratio, also to compare mu follower with CCS at a high output ratio.

It should work in switching mode but is not happy at low idle currents and likes to oscillate. I need to experiment with that some more.
 
Last edited:
Correct. A quick check in a russian 6C45 Ua/Ia shows ideal, symmetric swing, low 2H conditions with a flat loadline(CCS/no load). As soon as it gets downloaded symmetry fails severely. 10k should give at least 2% 2H at full swing. What Ua/Ia-curves do you use?

heh heh ...

I get 2% 2H at somewhere around 2W output with the 10K load and 12mA idle current. But this is baseline to look at AT vs CCS. I have curves below.

the fft plots are 10mW, 100mW, 250mW, 500mW, and 1W fundamental power, zero db is 1W

I only show these to 1W because my interface adds too much H3 above about 3 volts input and I don't have my scaler set up yet. Anyway it's a baseline from which to compare CCS and mu follower, and various current ratios.
 

Attachments

  • impmult-1W.png
    impmult-1W.png
    100.7 KB · Views: 56
  • impmult-500mW.png
    impmult-500mW.png
    99.3 KB · Views: 62
  • impmult-250mW.png
    impmult-250mW.png
    99.1 KB · Views: 64
  • impmult-100mW.png
    impmult-100mW.png
    97.2 KB · Views: 86
  • impmult-10mW.png
    impmult-10mW.png
    97.4 KB · Views: 391
  • 6C45P-loadline.jpg
    6C45P-loadline.jpg
    89.9 KB · Views: 408
Last edited:
heh heh ...

I get 2% 2H at somewhere around 2W output with the 10K load and 12mA idle current. But this is baseline to look at AT vs CCS. I have curves below.

the fft plots are 10mW, 100mW, 250mW, 500mW, and 1W fundamental power, zero db is 1W

I only show these to 1W because my interface adds too much H3 above about 3 volts input and I don't have my scaler set up yet. Anyway it's a baseline from which to compare CCS and mu follower, and various current ratios.

Oops, that load line chart is for a WE437A; but the 2 charts I have for the 6C45P are pretty much the same as that one, within normal vacuum tube variations. Plate resistance is more like 2K at this OP giving a damping factor of ~5

The big idea I'm investigating here is that the tube interacts with the loudspeaker load as does a parafeed SET output stage, rather than the typical voltage follower based hybrid which isolates the tube from the loudspeaker load.

I can configure this either as impedance multiplier or follower, with CCS, anti triode or mu-follower tube stage.
 
Last edited:
I measured the dynamic resistance of the 2.5V zener diode I was using for cathode bias of the 6C45P and found it was approaching 50 ohms depending on current, and was adding some 2nd and 3rd harmonic distortion due to it's nonlinear resistance with current. I replaced it with a TL431 and the THD at one watt went from 2% to 1.6%. I also rescaled the analyzer to 10W output and remeasured at 0.1, 1, and 10 watts.
 

Attachments

  • impmult-TL431-10W.png
    impmult-TL431-10W.png
    115 KB · Views: 187
  • impmult-TL431-1W.png
    impmult-TL431-1W.png
    99.5 KB · Views: 203
  • impmult-TL431-100mW.png
    impmult-TL431-100mW.png
    97.3 KB · Views: 204
I set up the impedance multiplier amp for CCS operation on the triode plate and a 46:1 impedance multiplier. This puts about 23K load on the 6C45P. At the OP of 10mA, 175V the Rp is about 2K-2.5K so the current damping factor is about 10.

As expected, this reduces the distortion across the board and preserves the predominant 2nd harmonic until clipping, which is about 15W. THe power level is relative to 10W (0 dB) so the plots are 100mW, 1W, 2.5W, 5W, 10W, and onset of clipping at 15W. There is a soft clip action as the tube saturates and the CCS tops out on the reflected speaker load. I have the amp set up as a 10W amp so am pretty happy with the power margin.

I'm going to build the 2nd channel and play it in stereo
 

Attachments

  • impmult-100mW-CCS22K.png
    impmult-100mW-CCS22K.png
    90.3 KB · Views: 189
  • impmult-1W-CCS22K.png
    impmult-1W-CCS22K.png
    92.8 KB · Views: 173
  • impmult-2.5W-CCS22K.png
    impmult-2.5W-CCS22K.png
    92.1 KB · Views: 38
  • impmult-5W-CCS22K.png
    impmult-5W-CCS22K.png
    94.6 KB · Views: 36
  • impmult-10W-CCS22K.png
    impmult-10W-CCS22K.png
    96.8 KB · Views: 43
  • impmult-clip-CCS22K.png
    impmult-clip-CCS22K.png
    106 KB · Views: 45
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.