Tuning an amp with an O'scope

[Fischerman]

I have a homebrew amp that started out as a power scaled Express using a Ceriatone kit. It never sounded right...there was some buzzy/fizzy stuff on top of the notes and it didn't sustain right. I tried a bunch of different preamps (Express, K60, Concorde, JCM800, Bogner XTC Blue, etc.) and none of them sounded right. So I ripped out the only remaining Ceriatone board (the preamp) and rebuilt a new one using G10/eyelets/and a Concorde preamp. It still has the buzz/fizz and won't sustain right. I've tried lots of tweaks but recently put it back to the 'original' Concorde circuit. I'm pretty sure it's in the preamp/PI because the problem is still there when the power tubers are running clean (remember, this amp has power scaling...so it's possible to run the preamp/PI dirty while the power tubes are still clean). When the power tubes are dirty they mask it a little bit (sounds better) but it still ain't right.

I have a O'scope and a signal generator but I'm not sure how to use it to tune the amp. When cranked (or anywhere close to it), the waveform at the input grid of the PI is round on the bottom half and chopped off on the top half but there is a spkie/overshoot on the 'tail end' (right side) of the chopped-off top half of the waveform. There is a series 220k resistor just before that PI input grid and the signal is the same there except for no spike/overshoot (it appears to me that the signal is 'correct' here because the top half of the waveform is clipped and the bottom half is not and there is no spike/overshoot).

The signal on the PI cathode and the other grid both have this spike/overshoot but even worse.

I've tried putting pots in place of the tail and PI cathode resistors and couldn't really help it much (a slightly larger tail up to ~22k helped then didn't do much after that...a slightly larger cathode R up to just below 820 seemed to help a tiny, tiny bit but then got worse and the amp lost some balls after that).

[John_P_WI]

Fischerman,

It's late so I'll be brief and post more at a later time. I use a .1 volt peak to peak sine wave signal at 1000 hz for testing. Starting with the 1st stage look at the signal shape, move to the next stage, how does it look? You will notice easily if the top is cut off (V + supply limited) or the bottom cut off (change cathode resistor value). Play with V+, plate R, or cathode R to "center" the signal swing. Remember the signal will be inverted too. If the signal is too hot and is slamming against the V+ rail, change the signal voltage divider to dump some of the signal before the next tube grid. If you slam a tube with too big of a signal, you will get a terrible "sticky" type sounding distortion. I have found that a grid block resistor on the next stage can also help a previous stage stay linear, if you can't figure out why the previous one is clipping.

I try to tune for symetric clipping, that is the top and bottom maintain a somewhat similar wave form.

The spike that you mentioned was most likely a transient when one of the stages was grosly overdriven past its linearity. BTW, it is amazing how much distortion content is contained in a signal that still looks like a sine wave. I think it is harder to make a good sounding distortion amp than a clean amp.

Good luck,

John

[Fischerman]

Thanks John.

I have been using the same signal; 100mV @ 1KHz. I had sort of done what you describe although I don't really know what to look for.

The signal off the 1st stage plate is clean (looks like a perfect sine wave). The signal off the second stage plate is clean. Then it goes into a direct-coupled gain stage/cathode follower pair...I don't remember what the signal looks like at the third stage grid. But at the cathode of the cathode follower the signal is clipped on top and round on the bottom...doesn't look like the bottom half is clipped at all. Should I be getting some clipping on the bottom half? I thought that a GS/CF pair clipped somewhat symmetrically (but that's based on my experiences with the GS/CF pair in a Marshall or Hot Rodded Marshall...where there is a tone stack after the CF).

After the CF it goes straight into the .022uF PI input cap but there is a 220K series resistor between this PI input cap and the 1M grid load. So on one side of that PI input cap (the CF side) I have that signal that's clipped on top and round on the bottom...but on the other side of that 220k resistor (the PI grid side), there is almost the same waveform except for that 'spike/overshoot' at the trailing edge of the clipped top-half of the waveform. I tried a 56K grid stopper right on the PI socket but it didn't seem to help.

I'll go through the process again and write everything down and see if something sticks out. Thanks again...it seems so close...it seems if I could just fix one problem everything else would fall into place.

[Sonny ReVerb]

Silvatone's got a very nice page with waveforms at various points in the circuit of a November:

http://www.silvatone.bravepages.com/Nov ... hnical.htm

The November does have a cathode follower so it may give you an idea of what the signal should look like.

[UR12]

I know this may sound a little picky but I shy away from the 1khz tone. It's great for testing transformers and speakers but I use 440hz and some of the 220 and 880 octave above and below a above middle c. Some amps, especially Marshalls, have a big dip/scoop in the mids and if you use 1k to tune these amps they may not come out sounding like you wanted. Guitars don’t normally produce fundamental notes below say 80hz and can go to 880. Of course 2nd and 4th harmonics can get them anywhere between and close to the 4khz range. Depending on your tone controls or coupling cap/grid resistor values you will usually get a 3db- 6db roll off per octave. This is the frequency ranges I like to play with. All of these can be verified with an oscilloscope.

I have read in a few places that clipping the bottom half of the waveform produces more even harmonics than clipping the top. If you can find an operating point for the pres (check out the tube characteristic curves in the tube manuals) that helps to clip the bottoms then you may get a creamier/ warmer sounding distortion. Phases get inverted after each stage in an amp (Unless you have a CF or noninverting stage where the output of the stage comes off of the cathode) so you need to compare the phase of the input to the phase of the output to determine whether you are clipping top or bottom of the waveform. You can change the operating point by changing the cathode resistor, plate resistor or changing the power supply voltage going to the tube. You could even go with an adjustable negative bias voltage on your preamp tubes to set the operating point anywhere you want just like you do on your power tubes if you wanted to get elaborate.

[John_P_WI]

Fischerman,

Forgot to mention, using a PPIMV (even temporary) will allow you to listen to the tone while adjusting drive / volume and will help sort out pre-amp issues. It is easy to hear a change and probe with a scope and see what is going crazy. Similary pots can be used to tune (dump) the signal and resistor / cap decade boxes can be used to tune the gain from the cathode pre-amp stages.

I have found that the cathode follower stages can be a little more difficult to set-up and I usually dump a lot of the signal leading up to the cathode follower to keep it from railing.

Distortion is like a favorite color, eveyrone has a different one. I like smooth and creamy, not harsh and buzzy which comes from the spikey overshoots.

John

[Fischerman]

Thanks for all the responses.

I'll try some lower frequency signals and see what I get. I always wondered about this because it's mostly low notes, struck hard that cause the problems (or at least...that's when they are at their worst).

This amp already has a post-PI MV...it's a necessary part of power scaling if you only scale the power tubes (which is how I'm running it).

I like smooth and creamy too but I don't really expect to get that out of this amp...it seems that the Express/Komet/Concorde lineage just naturally have a more aggressive distortion. But it shouldn't have that buzz/fizz on top and it should sustain for two days. After I button this one up I'm about to start my first ODS prototype...for the smooth and creamy (then I'm sure I'll be going through this whole gyration again).

[Fischerman]

Wasn't able to spend much time on it last night...but I did go through the preamp with a 100mV @ 1kHz signal again.

First stage is clean.

Second stage grid is clean up to about 6.5 on volume, then after that top half gets more fat and rounded. At max volume top half is noticably more fat/rounded and has a noticeable 'lean' to the right at the tip. These changes happen fairly gradually as volume is increased.

Second stage plate is similar to the second stage grid except a bit worse and the bottom half gets slightly more rounded at max volume....again everything happens gradually as volume is increased.

Signal on 'other' side of coupling cap after second stage plate is totally inverted from the second stage plate.

Signal on 3rd stage grid is where I suspect things go south. At about volume on 2, top half gets clipped/flattened and won't go any higher in amplitude...just flatter and flatter. At max volume, top is fairly square but the bottom looks like an upside down sharkfin. Put another way, the transition from top half to bottom half is almost a vertical line on one side and more sinusoidal on the other.

3rd stage plate (which is direct-coupled to the CF grid) has a similar top half as the grid, i.e. top quickly gets flat and won't increase in amplitude...just gets flatter and flatter. But the bottom is different than the grid...no upside down sharkfin but more similar to the top half except not quite as square (has rounded corners where the top half has fairly sharp corners). At max volume it's fairly square except the rounded corners on the bottom half. At volume ~6.5, the top is square, bottom is rounded, and the 'sides' are vertical (this is the setting I was referring to in the op).

Cathode follower output is pretty much identical to the 3rd stage plate. Doesn't look like the CF is distorting at all.

[bcook]

Signal on 'other' side of coupling cap after second stage plate is totally inverted from the second stage plate.

Am I reading this correctly, when the 2nd stage plate signal is positive, the 3rd stage grid signal is negative?

[Fischerman]

Am I reading this correctly, when the 2nd stage plate signal is positive, the 3rd stage grid signal is negative?

Pretty much...or at least it appears that way. When I probe the plate/coupling cap junction...I get the signal described above as 'Second stage plate signal'. When I probe the other side of that coupling cap (which also junctions with a 330K to ground and 200K grid stopper mounted on socket), I get a mirror image...looks like it was flipped upside down...signals look so similar as to look exactly the same (amplitude included)...just inverted.

However, the signal on the 3rd stage grid is separated from the coupling cap by that 200K grid stopper...and the signal there is described as 'Third stage grid' above...it is quite different from the signal on the coupling cap. This is where the upside down sharkfin appears.

[Fischerman]

I made a mistake earlier...even with my notes I still described this wrong:

When I probe the plate/coupling cap junction...I get the signal described above as 'Second stage plate signal'. When I probe the other side of that coupling cap (which also junctions with a 330K to ground and 200K grid stopper mounted on socket), I get a mirror image...looks like it was flipped upside down...signals look so similar as to look exactly the same (amplitude included)...just inverted.

It doesn't look like it was flipped upside down...it looks like it was flipped left-to-right (like a mirror image). The 'lean to the right at the tip of the top waveform' becomes a lean to the left at the tip of the top waveform.

I tried 440Hz and 880Hz last night. 440Hz doesn't give much on the screen to look at but 880Hz worked nicely. One thing I tried (and 'scoped but haven't listened to yet) was to attenuate the signal going to the 3rd stage grid by adding a 220k series resistor between the coupling cap and the 330k-to-ground/200k grid stopper junction (a voltage divider like the K60 has except I have added the 200K grid stopper). The interesting thing was...even though I cut the signal going into the 3rd stage by almost half...the CF output AC voltage didn't change...it's still putting out about 54vac at max volume (which is a little more than the PI plates!).

[skyboltone]

I tried 440Hz and 880Hz last night. 440Hz doesn't give much on the screen to look at but 880Hz worked nicely.

Change the time base and voltage scale!

[Darkbluemurder]

Am I reading this correctly, when the 2nd stage plate signal is positive, the 3rd stage grid signal is negative?

Pretty much...or at least it appears that way. When I probe the plate/coupling cap junction...I get the signal described above as 'Second stage plate signal'. When I probe the other side of that coupling cap (which also junctions with a 330K to ground and 200K grid stopper mounted on socket), I get a mirror image...looks like it was flipped upside down...signals look so similar as to look exactly the same (amplitude included)...just inverted.

However, the signal on the 3rd stage grid is separated from the coupling cap by that 200K grid stopper...and the signal there is described as 'Third stage grid' above...it is quite different from the signal on the coupling cap. This is where the upside down sharkfin appears.

This doesn't sound right to me. Are you sure the coupling cap is not leaky?

[John_P_WI]

Hello Fischerman,

My guess is that you are wildly overdriving the 3rd stage. Try disconnecting the wire to the grid of the 3rd stage and view the wave form at the V2 coupling cap. If this helps, then knock down the signal before the 3rd stage and take some gain out of 3rd stage too (increase Rk). Even a 3 volt peak to peak signal thrown into a gain stage of 20 or 30 will try to swing 60 or 90 volts, most likely causing the tube to drive way past its linearity creating your problems.

John

[Fischerman]

Yea, I think it was the 3rd stage getting way too much. I cut it way back and it does sound better but I just can't seem to get this amp sounding satisfactory to me. After the coupling cap after the second stage: I put two 470k in series with a 500pF across one of them, then a 470k to ground, then to the grid (no grid stopper). This helped a lot but I still am not digging this amp.

The coupling cap is not leaky, I made a mistake in my earlier description...it's not inverted (as in upside down) on the other side of that cap...it's backwards (as in a mirror image).