[dude_500]

I'm well on my way to completing a blue hawaii build. I have both channels built and am testing them. They are built according to the original schematic, with the change of using a depletion-mode FET current source. The board is approximately the same layout as the board files posted to Kevin Gilmore's site. The input J-FET's are matched 2SK170.
 
They seem to work great at low input signal level, but start do a strange clipping behavior on the low side of the wave as the input goes up (also the top of the waveform becomes slightly distorted even before clipping at the rail, but nothing like the bottom of the wave).
 
Attached are pictures sequentially increasing signal level. The dashed lines indicate approximately the +-400V lines
 
Both channels do this in the same manner. Is this perhaps actually expected behavior? I was under the impression it was a nearly rail to rail amplifier. Any idea what's going on here? The cathode is biased at -350V as expected.
 

 

 

 

 

 

[NA Blur]

Clipping is always a concern with amplifiers.  There is going to be some point where the amp cannot give any more voltage and the signal clips.  Just about every amp and power supply for that matter that I have had my hands on clips at some point.  You cannot expect the thing to have infinite gain especially if you limit the power in some way.
 
A quick wiki would have shown this:
http://en.wikipedia.org/wiki/Clipping_%28audio%29

 

[dude_500]

Quote:

Clipping is always a concern with amplifiers.  There is going to be some point where the amp cannot give any more voltage and the signal clips.  Just about every amp and power supply for that matter that I have had my hands on clips at some point.  You cannot expect the thing to have infinite gain especially if you limit the power in some way.
 
A quick wiki would have shown this:
http://en.wikipedia.org/wiki/Clipping_%28audio%29

 
I'm well aware of what clipping is and why it occurs in amplifiers. The point here, is the amplifier has a -400V rail and starts clipping on the low-side a little below -200V.

 

[kevin gilmore]

The Rp of the output tube limits the maximum negative swing to about -350V.
Some of the tubes i have tested have much lower Rp and you can get to about -375V.
This is normal.

 

[dude_500]

Quote:

The Rp of the output tube limits the maximum negative swing to about -350V.
Some of the tubes i have tested have much lower Rp and you can get to about -375V.
This is normal.

 
How do you get 1500v p-p output out of the amplifier then? I can only get just over 800v, since it starts clipping at -225V (150V/div on my traces).

 

[kevin gilmore]

You can put a scope probe on the cathode of the tube, and you can determine when the tube is hard on.
You may have the current source too high, it should be about 17ma. Assuming you are using the 10m90s
there should be no trouble there.

 

[dude_500]

Quote:

You can put a scope probe on the cathode of the tube, and you can determine when the tube is hard on.
You may have the current source too high, it should be about 17ma. Assuming you are using the 10m90s
there should be no trouble there.

Scoped the cathode, it is clipping to -400V when the output clips at ~-200V output. So it almost seems like the output needs more gain for the same cathode voltage. Is there some way to accomplish this? I'm not seeing any way, since the screen and suppressor resistors aren't there to set gain (as far as I know). Adjusting the anode current doesn't change much of anything. Pulled it down to 17ma, no change. Changing anode current while it's on doesn't change output at all, which surprised me a bit.
 
Are my tubes just bad and lack sufficient gain? They tested above spec according to the ebay ad.
 
In this picture: Cathode is the bottom trace, clipping trace is the output (and clean one is the input). Again, the dashed line is the -400V rail, showing that cathode is clipping.
 

 

[kevin gilmore]

I measured my original prototype last night with 10 year old richardson electronic cheap tubes, and the minus clips at -340v.
 
If the cathode is at -400v, and G1 is at -400v, then the tube absolutely has to be hard on.
 
So...
 
1) Filament voltage is way low  needs to be 6.3V at the tube.
2) Tubes are junk
3) for whatever reason the tube has really low gain.
4) current source has a protection zener that is limiting voltage swing. (i don't know
of any depletion mode fets that could possibly do this, certainly the enhancement
mode fets have protection zeners in them)
 
You can rewire the 470 ohm resistor going to G2 from the plate of the tube
to +400v to increase the voltage gain.
 
The Rp of the tube should be less than 4kohm, so there should be no more than
50 volts across the tube.

 

[dude_500]

Thanks for doing the measurement. I took a closer look at the cathode voltage by taking a differential measurement across the 2SJ79, and found the cathode isn't truly reaching negative HV bus when it's saturated. It's actually about 7-8 volts shy of it. If I keep cranking the input signal up, it then reaches truly putting negative HV on the cathode, and it clips at about -360V (as shown in the 4th picture of my original post).
 
Here are 2SJ79 voltage drop traces, 20V/div. Center of scope is 0V.
 

 
I don't see any reason why the 2SJ79's shouldn't be able to easily reach practically 0V in their standard operation range in this amp. Do you think maybe the 2SA1156 Q10/Q16 current sources need to be turned up to give Q13/Q14 a little bit more voltage swing to drive Q8/Q15 more?

 

[spritzer]

Are you using untested tubes off ebay????  Are you mad?  Get properly matched tubes from a reliable seller and Kevin's point about the filament voltage being too low is something to look for. 

 

[dude_500]

Quote:

Are you using untested tubes off ebay????  Are you mad?  Get properly matched tubes from a reliable seller and Kevin's point about the filament voltage being too low is something to look for. 

I am using a matched quad that tested great off ebay. I merely point out that it could be the tubes, since I don't have a tube tester myself, and haven't manually tested the tubes.
 
Filaments are powered by a regulated 6.3V DC power supply, and draw the expected current.
 
But anyways, I'm pretty sure the problem is that Q8/Q15 aren't pulling the cathode low enough as mentioned in my last post. Just need to figure out how to fix that.

 

[spritzer]

I've bought enough tubes off ebay to never trust the sellers and test every tube.  I've also built enough Blue Hawaii's to be highly suspect of the tubes here. 

 

[NA Blur]

Wow thanks for this great post all.  Even though we all expected it to clip at some point, clipping that early was a surprise.
 
When I run power supplies in various laser systems I build using an RLC type lamp pulse, the clipping is almost always related to the efficiency of the PSU and not the RLC components.  Can you better cool the amp to see if the cause is thermal?
 
The next thing to try would be higher quality tubes and test them.

 

[dude_500]

Put together a quick tube tester, they seem to be just fine.
 
A few more things I've found:
 
*Above 10khz, there is substantial output distortion, and the output to a sinusoidal input begins to approach a triangle. By 100khz, a low amplitude nearly perfect triangle is all that's left for a clean sine input of the same amplitude that made lots of nice output at <10khz. Basically, it looks like it has a maximum slew rate of about 9V/uS.
 
*There is audible humming coming from the amplifier for very narrow bands around 5khz, and 10khz. I'm not sure what's humming. I've tried putting an insulating stick to my ear and poking things which generally picks up most noises pretty well, but I haven't been able to reliably track down the noise source.
 
 
Both issues seem to be all the way back at the input stage to some degree, of course this doesn't mean a whole lot since there is feedback. This makes it difficult to track down the source of the problem, though.
 
 
Also, I'm using IXTP08N100D2 current sources with 100 ohm gate stoppers. I replaced these with 50k ohm power resistors though, and both problems still existed nearly identically, so I don't think there's any trouble there.

 

[dude_500]

There's got to be something about this amplifier that I don't understand. Even simulation agrees that it should clip around -200V! In order to get down to -350V or better, it would need a grid voltage of truly near zero, not -5 or -10 volts. But the way I see it, is the grid voltage can't approach zero, since its minimum magnitude should be the sum of the threshold voltage of Q8, and the bias voltage due to the current flowing through P2, the 1K trimpot (which is a couple of volts). The sum of these would be in the range of 5 to 10 volts, which is the minimum grid voltage I'm achieving.
 
So can someone explain how the grid is supposed to approach zero?
 
Below is a simulation that does precisely what my amplifier does, cathode clips before grid voltage reaches zero. The only solution I see is to offset the grid a little bit, but I must be missing something about how the circuit works.