Smartest way to 200W?

[Gaz]

First of all, it would be good to confirm where it is running out of headroom first - at the PI or the power amp. If you have a dual trace scope you can scope the output of the PI and the power amp simultaneoulsy. As you turn the volume up if the output of the power amp clips while the PI is still clean then the headroom problem is in the power amp. If they distort at the same time and the two wave forms are identical in shape then it's likely running out of headroom at the PI.

Is there not a dropping resistor in your power supply between the screen and PI nodes? If so then raising the supply voltage to the PI is simply a matter of changing that resistor for one of less resistance. Changing that one resistor will of course also raise the voltages downstream in the pre-amp. If you wanted them to be unaffected you'd have to change the resistor after the PI node to one of higher value. If you determined the headroom problem is not at your PI then there's no reason to do any of this of course.

Hope this helps,

Greg

Greg, it is beginning to clip before the power amp, but the out actually hard clips before PI. Lowering the tail resistor as es345 suggested did give the PI some more headroom, and now the output clearly clips before the PI.

Unfortunately, it's not giving me a single watt more output. Actually I'm measuring slightly more output power with the original higher tail resistor. I think what I'm actually observing is that the power amp clips more 'softly' when the PI rounds off first, which make is seem like I'm getting more power. With the lower value, and the output clipping first, the output clips more abruptly, and it's easier to see on the scope. Does that make sense?

[GregS]

Unfortunately, it's not giving me a single watt more output. Actually I'm measuring slightly more output power with the original higher tail resistor. I think what I'm actually observing is that the power amp clips more 'softly' when the PI rounds off first, which make is seem like I'm getting more power. With the lower value, and the output clipping first, the output clips more abruptly, and it's easier to see on the scope. Does that make sense?

If it's the PI that runs out of headroom first then the power amp isn't actually clipping (or running out of headroom) but simply amplifying/copying the rounded signal output from the PI. Power amp clipping does clip harder and more abruptly and tends to be more symetrical (at least in the amps I've observed) so can be distinguished from the triode type distortion from the PI.

Anyway, from what you describe, it sounds like your power output is limited to what it is by the power amp. Again, it seems to me that your results should be expected from the relatively low voltages being supplied to your power amp - and that bringing the voltages up to the same spec as the Hiwatt would increase your output power significantly. I realize it's a big deal to replace the power transformer both expense-wise and effort-wise, and I'd hate if you did that based on my input and it didn't help much. If you're not finding the headroom to be a problem in actual use then it may be a better idea to leave it as is.

Greg

[Gaz]

If it's the PI that runs out of headroom first then the power amp isn't actually clipping (or running out of headroom) but simply amplifying/copying the rounded signal output from the PI. Power amp clipping does clip harder and more abruptly and tends to be more symetrical (at least in the amps I've observed) so can be distinguished from the triode type distortion from the PI.

Anyway, from what you describe, it sounds like your power output is limited to what it is by the power amp. Again, it seems to me that your results should be expected from the relatively low voltages being supplied to your power amp - and that bringing the voltages up to the same spec as the Hiwatt would increase your output power significantly. I realize it's a big deal to replace the power transformer both expense-wise and effort-wise, and I'd hate if you did that based on my input and it didn't help much. If you're not finding the headroom to be a problem in actual use then it may be a better idea to leave it as is.

Greg

Greg, that's a good way of putting it, that the output wave form is merely copying the PI. Thanks for the thoughts - still not sure if I'll spluge for the custom PT

[es345]

If you don't want to go for a new PT you have 2 cheaper choices which might help (perhaps even more)

1. Test the output power after increasing the value of the filter caps, e.g change from 2x 100uF to 2x 220uF or 2X330uF.
Change for this test your rectifier diodes so that they are able to handle the current during switch on.
As a result the voltage HT1 and HT 2 at full power should be higher at full load > more output power

2. Go for AB2 mode: insert MOSFET follower circuits between PI and Power stage (you will find several designs here in the forum).

[trobbins]

Yes it is very wise to at least use a NTC thermistor in the AC primary, or a delay/soft-start, as the power rating starts to push up the filter capacitance level, even when you have a grunty diode bridge.

[Steve Conner]

A couple of points:

If your power tubes are clipping by running into grid current, then they will clip the positive excursions at the PI plates too. The PI may look as if it's running out of headroom when it's not.

Going to Class AB2, you have to worry about control grid dissipation as well as the screen grids. I don't think any modern audio tube is specified for any control grid dissipation at all.

[es345]

A couple of points:


Going to Class AB2, you have to worry about control grid dissipation as well as the screen grids. I don't think any modern audio tube is specified for any control grid dissipation at all.

In the attachement ther is one example to get the thing it under control.
Looking to the upper path: ZD3 limits the maximum positive voltage, R9a generates a little bit of smooth clipping.

The value of ZD3 depends on the power tube and power section which is implemented.
The way I have used define the value is to limit the G2 dissipation just below design maximum when having about 30% clipping at powerstage. The value shown is used with GU50 at 850V Ua and 300V Ug2.

Important:
this method is applicable for bassamps as you don't go into heavy overdrive with them.

Regarding control grid dissipation indeed there are no specs for modern tubes - unfortunately.
For tubes like KT88, KT120 I would estimate between 0,5 -1 W

[exclamationmark]

How do they actually get that amount of power out of only 4 x KT88!? Considering that the maximum efficiency of a class AB amplifier is somewhere around 80% (and realistically only 65-70%) of plate dissipation wouldn't they be absolutely flogging those tubes at max power?

[trobbins]

My 2x KT88 VASE amp is relative clean to about 75W and exceeds 100W when overdriven. It's got a doubler that generates 500V at idle, and has to push through a 65W rated OT. They idle at 22W, and although I haven't done an exact power loss measurement when cranked, the average dissipation doesn't go up much using nominal power loss assessment.

[imaradiostar]

How do they actually get that amount of power out of only 4 x KT88!? Considering that the maximum efficiency of a class AB amplifier is somewhere around 80% (and realistically only 65-70%) of plate dissipation wouldn't they be absolutely flogging those tubes at max power?

I read a review of the big Reeves bass amp somewhere and they claimed the plate voltage was well over 700 volts and the screens were above 400 volts. This could have a lot to do with it. That amp tested in excess of its rated 225 watts. I've read that the old hiwatts had similarly high voltages.

If each output valve is rated for 55 watts, it can readily dissipate 100 watts for a half cycle and close to nothing for the other half cycle. By extension, the theoretical limit for a class B amp made up of 4 of that same valve should be over 400 watts. In the real world it would tend to blow up, but over 200 should be very doable with high enough plate voltage.

This same logic is applied on the EL34 datasheet where a pair can produce 100 watts in class B. 100 grungy intermodulation distorted watts...from two EL34's. It would probably have worked great for an outdoor PA that was voice-only, or other less sonically demanding applications.

Jamie

[es345]

I read a review of the big Reeves bass amp somewhere and they claimed the plate voltage was well over 700 volts and the screens were above 400 volts. This could have a lot to do with it. That amp tested in excess of its rated 225 watts. I've read that the old hiwatts had similarly high voltages.

If each output valve is rated for 55 watts, it can readily dissipate 100 watts for a half cycle and close to nothing for the other half cycle. By extension, the theoretical limit for a class B amp made up of 4 of that same valve should be over 400 watts. In the real world it would tend to blow up, but over 200 should be very doable with high enough plate voltage.

Jamie

To make that visible some results out of an estimation tool I have written for myself.


- Case 1a is showing a push pull 4xKt88 config at 600 V Ub (at full power) showing about 200W output. The curve on the right side is showing the plate dissipation in relation to the % of drive. 100% drive is max output without clipping.

- Case 1b is showing the instantaneous power during one sinus curve, paramater is the % of drive.

- Case 2 is showing the same config as in case 1 but with a higher winding resistance at the primary (100Ohm instead of 40 Ohm in case 1). The power is now lower (165W).

So a good OT with low resistance at primary clearly helps to get more output (this might be obvious, but sometimes it is forgotten).

Hans- Georg

[exclamationmark]

Interesting diagrams! The average dissipation part was the missing link in my understanding - I've played around with some sims and it's quite evident that if you bias heavily towards class B you can send the load line almost vertical into the dissipation limit and still be within acceptable limits.

[imaradiostar]

To make that visible some results out of an estimation tool I have written for myself.

Nice diagrams. I have an earlier version of your spreadsheet and I've found it to be a great tool. Would you mind sharing the current version? It looks like you've added some helpful improvements!

Jamie

[es345]

Here is the updated spreadsheet

I have now taken into account the resistance of the primary coil to calculate the plate disspation and the output power.
I have found that the estimated value tends to be to big (obviously).

Please keep in mind that it an estimation aid only.
The idea behind is that in a well designed powerstage there should be a sinus at the anode of the powerpentode when driven with a sinus.
The OT is assumed to be linear
Cells marked green in the summary sheet : (reasonable) input data, no boundary checkings

Hans- Georg

[imaradiostar]

Thanks, this is a neat tool!

When you say "too big" do you mean the spreadsheet's power estimate is too high? What about plate and screen current?

I put some known circuits into the spreadsheet and I'd say it was reasonably accurate. Of course, I have a table with DCR, various voltages and currents, and impedances from tubes and transformers I've used on my test rig to compare to.

It seems like the crucial values to make note of are how low the valve can pull the anode and the DC resistance of the transformer primary. I have a small amp that I built as a teenager. I assumed it to be about an 18 watt amp and eventually when I measured it two years ago I was surprised to see it only does about 13 watts RMS sine, regardless of which output tubes I put in it (it's set up for any normal 8 pin output). The output transformer, though sized for at least 25 watts, has some of the highest DCR primaries I've seen. I guess that explains a lot.

Have you found that it corresponds well with your own builds?

Jamie