Russian TrainWreck - Expresso Lite(Liverpool), Soviet tubes

[VacuumVoodoo]

well, what did I suggest yesterday

Get rid of the 5k pot and put a 47k resistor there instead.
C14: connect the low end to ground instead.

This has a chance of curing your headache.

[TheGimp]

I thought I had caught those Sunday when I corrected the feedback connection and changed the 120K back to 100K.

On a slightly different note, what switches are used for standby? I've not seen switches rated for 300Vdc+.

I may leave it out as I consider standby to be a relic of the days when the electrolytic caps were rated for operation, but not for the voltage stress that would occur before the output tubes started conducting. I've measured the B+ on a Marshall JCM-800 peaking well over the rating of the filter caps. Then again, even in operation the caps were marginal. Same with Traynor and other amps I've worked on.

One thing I considered was a Triac on the output of the diodes and before the first filter cap as the control device. Since charging current drops to zero off peak, there would be no problems shutting it off. In addition, by switching the pulstaing output of the rectifiers, the effective PIV is reduced to the peak value, not 2X the peak value. 600V triacs (MAC15-8) would probably be fine for up to about 500V B+.

Use a IL4116 or similar to drive it and take the power to drive the LED in the Opto from the Filiment/Heater supply. Since it is isolated the power can float with the heater, or even be biased above ground for hum reductino in the heater supply.

Has anyone tried this? Waste of tiem? Just use a 230V rated switch?

[PlinytheWelder]

Everyone in the biz seems to use a 240VAC rated switch without problems, I wouldn't worry too much...

[TheGimp]

Does anyone incorporate them (Ventilation Holes) in amps?

A lot of stereo amps have vents on the bottom and around the tubes and other strategic locations to keep component temp down.

I've not seen this with Guitar amps.

Why not?

Preliminary Layout, I need to figure out how to mount this before I go further.

I'm thinking four 10-32s with blind nuts ('T'-Nuts?)in the corners of the chassis. Screws up from the bottom.

Do the first two tubes need to be shielded? Or only the first one?

I don't see that the PI needs to as it does not have as much gain, and is dealing with high level signals.

Tks.

[TheGimp]

I've seen discussions on the advantages and disadvantages of solid vs stranded wire.

One big problem with stranded wire is it won't stay in place.

When I have trouble with stranded wires, I slide a short piece of shrinko (heat shrinkable tubing) over it, and tack it in place with a hot glue gun. If I need to remove it, it is easy to slit the shrinko, and pull the wire loose.

[M Fowler]

I like the way your amp is coming together refreshing to see some good home brews.

[TheGimp]

Thanks.

Wiring is complete, B+verified and Screen Voltage regulator adjusted to 245V.

I'll start checking the rest of the voltages and installing tubes this evening.

It ended up a tad crowded. A 13.5 X 6 X 2 would have worked out a lot better, but no such box exists so the 13.5 X 5 X 2 will have to do. I messed up a couple placements and had to re-drill mounting holes because of interference fits.

I still need to make up a new drawing of all hole positions.

Honey-do list has me pressure washing the porches today.

[TheGimp]

Carol Eugene, is up and running.

FB was inverted so I swapped the drive from the PI to the output tubes. Problem resolved. The input has a 1M resistsor althugh it is not in the photo. I ended up adding one at the tube as there is no grid bias resistor if no guitar is plugged in.

I left off one jumper. No biggie.

It' LOUD!

Also lots of hiss that increases with increasing volume setting.

Any suggestions on how to attack this?

Could an oscillation cause general hiss? I've seen it cause distortion. I'll look at the preamp and PI stages with the scope (40MHz) this evening.

On a brighter side, Hum is very low, especially when I put the bottom cover in place.

I forgot to scale the drive for only two 6P1Ps vs four EL84s, and that is probably part of the problem.

I had to set the speaker 12 feet away to cut down on it squealing like a stuck pig.

I am totally lacking in musical aptitude so I'm testing with a home made CBG. I'll get a friend to try it out once I get the hiss reigned in.

I really need to learn a little on a guitar, but after three attempts at lessons I'm hesitant to irritate my wife with attempts at what I feel is a lost cause.

[TheGimp]

A couple of observations.

(1) Cold solder joints are noisy. Especially on grid of V1a from the volume pot.

(2) Coax is your friend. The 5" run from the Vol pot to the grid of V1A was picking up noise.

(3) Ground Break (anti parallel 3A diodes in parallel with a 100R resistor) work on Guitar amps as well as stereo amps.

(4) DYMO stamps and plastic tape are available at Wally world.

(5) There is an unused triode in V2(b). Why not move V1A to this position, and parallel V1A and V1B as the first gain stage to reduce the noise. In addition, paralleling the two triodes reduces the output impedance driving the tone-stack. This seems to me to be a Win-Win situation. I implemented this change in my amp.

I still have more hum than I like, and need to continue working on it.

The cab will have to wait.

[TheGimp]

All controls at 12-Oclock except the vol at full resulted in 2Vrms noise into an 8 ohm speaker.

After replacing the vol pot, dressing leads, installing two grid-stop resistors, and going through 20 tubes (Voshod 6N2P-EB, 1977 through 1989) I've got it down to 0.190Vrms into 8 ohms.

With all controls at 12-Oclock, I measure 30mVrms noise into 8 ohms.

Still too much hiss, but the hum is much lower.

Chopsticks pushing wires had very little effect other than every time my hand got close hum went up, and when I pulled back it went back down.

Measurements were made with a Fluke 77 VOM and were probably not very accurate considering the bandwidth of the meter, but at least a good relative measurement so I can see when I'm making improvements that are less than I can hear.

[TheGimp]

I'm still at it.

I've gone through about 20 tubes, tweaked a couple of stages, and have the noise down to about 18mV rms with all controls at noon. With all controls at noon except vol at max I have 42mV noise + Hum with hum being a majority of that.

A friend loaned me a Fender Stratocaster (made in Mexico, not an older us made one) and I have a Weber VST Sig10F I've been testing with as a step up from the 12" Magnavox Stereo Pull I was using.

The amp is starting to tighten up. Hum is acceptably low (I found one input tube that was causing 72mV of it). Hiss is louder than I'd like but may be acceptable.

Using the Fender set at 5 with the non amplified pickup I can get clean tones up to about 5 or 6 and by the time I get to 7 it starts breaking up.

I'm getting 8W out with about 12 V sag when testing at 1000hZ with a tone test. I haven't tested frequency response yet.

I may take it to work tomorrow to send home with the friend for him to test it and see what he thinks.

[Colossal]

Gimp,

I've been enjoying your commentary and observations. Keep it coming.

Regards,
Dave

[TheGimp]

Thanks Dave.

I've had a lot f my questions answered by reading back through the list, but I can't read everything. Hopefully by posting here I'll be able to answer questions that some others may have.

It's certainly an interesting journey.

[TheGimp]

I'm still chasing my hiss issue. Roger (the friend who plays guitar, repairs guitars, and sells them) indicated it was finally down to a level that was acceptable for a guitar amp, but I'd like to reduce it further.

My feeling it that the 6N2P is noisier than the 12AX7, but I've read that 12AX7s in such high gain circuits are noisy and many people had/have issues with hiss.

My opinion (yea, I know they are like A.. H...s, everyone has one.) is that it should be possible to design a circuit that reduces noise and tube characteristic dependence which will allow for a wider selection of tubes to produce the same 'tone'.

Tone then becomes more dependent on circuit design (bias point) than tube selection. The downside of doing this is it reduces the stage gain and requires an additional gain stage, or changing the first stage from triode to pentode (EF86? or similar SQ tube?).

With that in mind I set out to do a little analysis

I used Trainwreck Express Voltage chart V1.1 for bias voltages to determine cathode currents (Vk/Rk=Ik). From this I plotted the bias point on the plate curve for the 12AX7 tubes.

I also measured the equivalent 6N2P voltages and calculated the corresponding current values.

Since I’m interested in minimizing noise (hiss in particular) the first three stages are the ones that are critical. Leaving the Phase Inverter (Splitter) but removing the first two tubes (three stages) results in a very quiet output. So I see no need for analysis of the PI on.


12AX7 analysis using the GE 12AX7 data sheet:

1) V1b – Input tube first gain stage.
Plate load - Tone-stack simulation indicates Zstack =136.7K (43.53mV/320.78nA)

Vk=1.5V Rk=1K5V yields Ik=1.00mA. This plots Va=182.5V
Ra = 100K||136.7K = 57.7K
ra = from the data sheet, I calculated 64.8K by interpolating between the values at 100V and 250V (data sheet, graphical method on curves =68.57K)
AV = 100*57.7K / (57.7K+64.8K) = 47.1 (graphical method yields Av=60)

Simulation yielded a gain value of - AV=43.72 Calculated and simulation agree, graphical does not.


2) V1a Tone stack buffer output amp
Va=210V and Vk=1.9 at Rk=2.7K yields Ik=0.704mA.
Vk=1.9V and Ik=0.704mA plots Va=210V
Ra = 100K||150K = 60K
ra = 67.2K (data sheet interpolation, graphical method on curves =68K)
AV = 100*60K / (60K+67.2K) = 47.16 (graphical method yields 52.5)

Sim – AV=37.1 Poor coorilation.


3) V2a PI Driver
Va=267V, Vk=2.8 and Rk=10K yields Ik=0.28mA.
Vk=2.8V and Ik=0.288mA plots Va=255V vs the measured 267 value.

Ra = 100K||1M = 90.9K (actually slightly higher due to bootstrapping from un-bypassed cathode resistor)
ra = 62K (data sheet, graphical method on curves =124K???)
AV = 100*90.9K / (90.9K+62K+10000(100+1)) = 7.816
(graphical method yields 36)

Sim – AV=7.79 Good agreement.

6N2P analysis:
The plot is not as detailed (poor resolution) as for the 12AX7.


1) V1b – Input tube first gain stage.
Tone-stack simulation indicates Zstack =136.7K (43.53mV/320.78nA)
Va=168V, Vk=1.547 and Rk=1.5K yields Ik=1.03mA.
Vk=1.547V and Ik=1.03mA plots Va=190V vs the measured 168 value.

Ra = 100K||136.7K = 57.7K

ra = 51.8K (data sheet, graphical method on curves =65K)
AV = 100*57.7K / (57.7K+51.8K) = 52.69 (graphical method yields 57.5)

Sim – AV=43.72 Sim is worse than calculated or graphical.


2) V1a Tone stack buffer output amp

Va=149.8V and Vk=1.47 at Rk=2.7K yields Ik=0.544mA.
Vk=1.9V and Ik=0.544mA plots Va=150V
Ra = 100K||150K = 60K
ra = 51.8K (data sheet, graphical method on curves =65K)
AV = 100*60K / (60K+51.8K) = 53.6 (graphical method yields 57.5)

Sim – AV=37.1 Whoa! Sim is way worse than calculated or graphical which are close to each other. But, sim is closer to 12AX7 Sim values. 6N2P taken from Duncan Amps 12AX7 and 12AX7 if from Koren.


3) V2a PI Driver

Va=210.6V, Vk=2.516 and Rk=10K yields Ik=0.2516mA.
Vk=2.5V and Ik=0.25mA plots Va=200V vs the measured 210 value.

Ra = 100K||1M = 90.9K (actually slightly higher due to bootstrapping from un-bypassed cathode resistor)
ra = 51.8K (data sheet, graphical method on curves =183K???)
AV = 100*90.9K / (90.9K+51.8K+10000(100+1)) = 7.886
(graphical method yields 36)

Sim – AV=7.79 Good agreement over all.

Now, what was this all about?

My plan is to reduce the gain of the first two stages with capacitively coupled Schade feedback, then add a LNA front end using either (1) an EF86 (or other SQ Pentode) or (2) a Low Noise sand amp.

This should allow me to reduce the noise in the two tube stages by using NFB, and compensate for the reduced gain with the new first low noise stage.

The advantage in addition to overall lower hiss is that the NFB in the first two stages minimizes the tube dependence and allows a wider spread of parameters while still hitting the same bias point. Ie, they become independent of tube selection. Bias point determines 'tone', not tube selection.

Does this make sense or am I just ‘off in the weeds’ so to speak?

end of part 1, to be continued

[TheGimp]

Part 2.

Since I don't have any EF86s I decided to try plan B first. I've got some low noise jfets and they seemed to be a good fit for a LNA front end.

I designed the JFET gain to be around 4, and cut the gain of the original first two stages in half each with plate to grid feedback.

This left me with the same gain going into the PI driver (gain of 7.7) and results in the overall amp gain remaining the same.

Why so little gain on the jFET? Why not more gain on it and less on the two tube stages? Because the bias on the first stage only allows 1.5V in or I drive it into grid conduction which I need to avoid.

Reducting the gain of the first two stages by 50% reduces the noise by a much greater degree than the gain reduction, so I'm not amplifying the noise as much.

Final result is that the hiss at full volume with the worst tubes (for hiss) I have is lower than it was before the mod with the best tubes I have. Background noise is now relatively independent of tube selection.

I loaned the amp out again last night and today the feedback was that it sounds like a 30W amp! It is LOUD and great sounding.

Distortion starts with the volume pot at noon, and the guitar at 5 which is where I understand it should occur. So it looks like this is going to work.

I will have to rewire the amp (and progbably make a new turret board) to clean it up as right now it looks like something a spider constructed with all t he added components just hanging in the air.