• 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.

Putting Stock In Carver?

BRSHiFi

Previously known as kingden
2008-08-28 4:27 am
Evanston, IL
I am messing around in Spice with DC clipper (restorer) circuit topologies. Carver claims they reduce crossover distortion and allows for cooler tube operation. I analyzed his circuit and see how a 6AL5 potentially pulls this off.

Since the 6AL5 is no longer in production, what type of diode do you recommend I substitute? My spice work so far yields no effectiveness. Does that circuit really work or am I not doing something right? My main motive is more efficient operation (less heat and longer tube life).
 

Attachments

  • clip.png
    clip.png
    21.7 KB · Views: 423
Last edited:
The diodes need to be reversed. The idea is for them to conduct on the opposite (negative) signal swing, at the same level as the internal U1 or U4 grid1/cathode effective diode does on the positive swing. That way the coupling capacitor does not charge up over many cycles. Both diode polarities
should present similar series resistance as well for symmetrical charging currents.
I've seen 6AL5 for $0.50 in quantity.
 
Last edited:
As I understand it, the DC restorer is a clamp circuit that prevents the output tubes from being biased too negatively on strong signals. This allows them to be operated at a lower idling current, to a point. The diodes in your circuit should be reversed and not be zener diodes. Carver uses regular 1N4148 switching diodes. And the normal negative bias voltage for the tube grids is seperate from the diodes source. Here is a link to Carver's own hand drawn modification of the Citation II.
 
Last edited:

BRSHiFi

Previously known as kingden
2008-08-28 4:27 am
Evanston, IL
I dropped in some 1N4148 diodes in spice. I cranked the bias way back and the distortion readings are hardly influenced, both with and without the clippers. Does that mean the circuit is a hoax or do I need to tweak for 6L6 usage? I figure less than 8 watts static plate dissipation with the clipper should be what Carver claims should work.
 

Attachments

  • Untitled.png
    Untitled.png
    38.3 KB · Views: 376
Last edited:
The diodes don't do anything unless the signal peaks exceed BiasB. Then the SS diodes and the internal tube diodes draw (hopefully) equal opposite currents, so that the coupling capacitor does not charge up and shift the DC operating bias level.

With no shift of DC operating bias for large signals, the design idle current of the tubes can be decreased, since they won't cut off from bias drift.
 
Last edited:

BRSHiFi

Previously known as kingden
2008-08-28 4:27 am
Evanston, IL
Why does carver boast the amps can be biased much cooler? In spice, with a continuous waveform, this does not happen without large amounts of notch distortion? The diodes are "supposed" to eliminate some of this. Hmmm.

I have the spice experiment attached.
 

Attachments

  • Simulation.zip
    421.5 KB · Views: 32
You seem to go for what Carver boasts about wirhot understanding what the circuit is supposed to do and under what circumstances, hence you are looking for the results of what he is boasting about in the wrong place. It only works when the driver attempts to drive the output grids positive, i.e. clipping. There is no change when the circuit operates normally so no use looking for it. 'Blocking distortion' is the term you need to look up...
 
I've been listening to a new PCB based amp using MosFet driven gates. The output tubes are 17JT6's, little 18watt sweep tubes. The MosFets prevent the conducting gate from adding an offset bias to the coupling caps when the gate starts going positive and conducting. I thought that was the objective of the diode also to clamp the gate from going more than .7v higher than the cathode to prevent the cap from taking a charge that effects the crossover distortion.
In previous amps, I've placed the diode from the gate to the cathode. I can't really say that it made a big deal, I do like the Mosfet drive!.
 

Attachments

  • P7070223_b.jpg
    P7070223_b.jpg
    155.8 KB · Views: 303

BRSHiFi

Previously known as kingden
2008-08-28 4:27 am
Evanston, IL
If carver's DC restorer circuit is moot when the amp operates normally.

How come the amplifiers boast distortion less than 0.5% despite the fact he claims mismatched tubes and very cool bias will work? Is it the feedback of the amplifier? After all his units have a switch to go between 12dB and 20dB of feedback.

Maybe we should not be putting stock into Carver's sales pitches about DC restoring?
 
Hmmm, I just looked at that Carver modded Cit II schematic ( http://www.vacuum-tube.eu/images/hk_citation2_mods/hk_citation2_mod2.jpg ) and it IS using a 91K Ohm resistor for the R3 (here), with just a 5 V drop in voltage below the bias V (55V from 50V).

That's a bit different. Should be causing some non-linearity by bringing the diode conduction in on any neg. signal peaks above 5V (AC). Apparently, it is not intended to hard limit the signal, but just provide 2X (versus the R24 100K in the bias path there) restoring current for the bias recovery on the coupling cap to be faster.

If one went further, and used several of these diode networks, each connected to say a 5V step further below the bias V, then they could progressively become deployed as the signal got bigger. Then one could bias the tube for class A op at small signals, and the diodes would push it toward class B bias as the signal got bigger. But a sudden drop of the signal back down to small levels would leave it in class B mode until the bias recovered on the coupling cap. That would remove most of the crossover notch then (for small signals), but likely cause some distortion in larger signals. Plenty of NFB available in the Cit II to fix that though. This would then be what is called a sliding bias scheme. Usual comments about them are that the noticeable time constant involved in the bias recovery can be heard to some extent. (Umm, except the diodes would need reversing to do this scheme. Carvers circuit is lowering the bias, ie, increasing cross-thru current as the signal gets bigger, ?? great way to melt down the output stage ?? what gives here ?)
 
Last edited:
There are two quite different reasons why output valves can shift to cooler bias when strong signals are present. Preventing or reducing this shift means that the valves can have cooler quiescent bias, which may reduce small signal distortion - the valves can start and remain at the right bias point instead of starting too hot.

The two mechanisms are:
1. Second order distortion in the valve. This effect is proportional to the square of the signal level so it operates at all levels but is only significant for stronger signals. Solution is fixed bias, or proper (e.g. zero crossing sampling) bias servo. CCS bias or quick'n'dirty servo bias make this worse, not better. Cathode resistor bias is a cheap and reliable compromise between these two extremes.
2. Grid current on positive peaks shifting the coupling cap voltage. Only occurs on peaks, when distortion will already be starting. Can cause blocking in worst cases. The Carver circuit, when properly adjusted, can reduce this problem but it does it by clipping peaks so you still get distortion. The full solution is DC coupling from a buffer stage or transformer drive.

Most hi-fi circuits are designed so that grid current does not happen, so Carver appears to be solving a problem which should never occur. However, when it does occur his circuit may help a bit by reducing blocking.

Adding feedback can make both issues worse, assuming that the OPT is a major limit on frequency response. At low and high frequencies, but not in the middle, the output stage needs more drive to offset the falling OPT response so more drive means more problems.
 
Here you can visual simulation of generic DC restorer.

DC Restoration
DC Restoration - Circuit Simulator

Scroll down to the chapter named "Clamper circuits".
Lessons In Electric Circuits -- Volume III (Semiconductors) - Chapter 3

I'll try to simulate this with LTSpice in PP tube circuit. Still not convinced it works as advertised.

@kingneb - if you are keen to decrease idle current of output tubes, you can try twin coupled McInstosh circuit. It is so old that copyright is expired anyway for whatever usage.
 
With only a 5V difference on the diodes midpoint versus the normal bias, it looks to me like Carvers circuit setup is hotting up the tubes with bigger signals until the signal gets big enough to start drawing grid currents to match the diode currents. This would seem to enhance crossover notch distortion at small signals (over biased) and take the tubes up to max dissipation (underbiased, or meltdown) at big signals. I can only describe this scenario as Cuckoo......???? Something must be wrong about that Cit II markup.

Unless the 5 V AC peaks is the maximum signal. Then it would act as a DC restorer.

I guess it could be seen as a class A circuit with sliding bias. Power conserving mainly. The tubes get hotted up with bigger signals to stay in class A. Assumes that you won't put a big sine wave into the thing, which would cause it to melt down the outputs. Blocking distortion (grid1 conduction), could then step in to stop the runaway if designed properly.
 
Last edited:
. . . CCS bias or quick'n'dirty servo bias make this worse, not better. Cathode resistor bias is a cheap and reliable compromise between these two extremes . . .
Yes ! The bypassed CCS's (one per tube) is the worst issue (I know, I did :cannotbe:).
In a 6BQ5 PP amp, the cathodes were rising from some 9..10 Volts nominal to more than 20 V at full power and horendous crossover distorsion occurs.

A way is to tie a zener in parallel with each CCS or cathode bias resistor.
Choose the zener voltage about 20% above nominal bias voltage.

Yves.
 
Last edited:

BRSHiFi

Previously known as kingden
2008-08-28 4:27 am
Evanston, IL
Bob claims that circuit is his secret weapon. The objective of this post was to evaluate the credibility and worthiness (the name of the post) of his work more than anything.

Analysis of clamper circuits does verify smoking-amp's description to be true and Carver's advertising "grey area."

Also, with a relatively small amount of feedback, does running tubes in pentode mode allow for distortion levels less than 0.3%. In spice, yes it does.