Solid State Guitar Amp Forum | DIY Guitar Amplifiers

Solid State Amplifiers => Tubes and Hybrids => Topic started by: ilyaa on March 26, 2014, 10:52:05 PM

Title: JMP 2100 not making power
Post by: ilyaa on March 26, 2014, 10:52:05 PM
so im trying to put some of my new skills to use -

someone gave me this amp to look at  - "sounds quiet"

B+/plate voltage - ~385VDC

tubes are EL34s
i did a quick bias, trying to get 14 watts or so dissipation.
started with one tube - looking for about 37mA. got it there with about ~-31VDC on the grids.
checked the other tube - it was pulling about 50mA.
switched them around to see if tubes or sockets and the readings moved with the tubes.
time for new tubes?

i then checked the power output with a 100mV sine wave into my (brand new) dummy load - it was giving me 12V p-p - so 4V RMS, so it's only making about 4 watts (into 4 ohms)?!?!?

this is a 50 watt amp...

only other thing that looked suspicious is the ripple on the main filter cap - about 8V p-p. is that acceptable? seems high....i thought the plate voltage seemed a bit low....

how am i doing?

(as far as i can tell the attached schematic corresponds to this amp...)

Title: Re: JMP 2100 not making power
Post by: Roly on March 27, 2014, 03:00:17 AM
Quote from: ilyaai then checked the power output with a 100mV sine wave into my (brand new) dummy load - it was giving me 12V p-p - so 4V RMS, so it's only making about 4 watts (into 4 ohms)?!?!?

The input signal level doesn't mean much; the important thing is that you measure the output power at the onset of clipping.  In this case it is likely to occur a lot earlier on one side than the other, and combined with overall low power output, and the lack of any other signs of trouble (i.e. the signal voltages coming out of the Phase Inverter are about the same as the bias voltage, so there is no lack of drive) then it is reasonable to suspect that the output valves have worn out and need to be replaced.  This is very common, and normally all you have to do.
Title: Re: JMP 2100 not making power
Post by: ilyaa on March 27, 2014, 10:56:24 AM
Quoteat the onset of clipping

yeah i turned it up until the sine wave started to wonk out a little bit.

the PI: you mean the AC voltages coming off the plates to feed the power tubes? these should be close to the DC bias voltages on the power tube grids?
Title: Re: JMP 2100 not making power
Post by: g1 on March 27, 2014, 12:05:45 PM
  Any chance you have an x10 scope probe?
The PI plate signals (peak to peak on scope) should be about double the bias voltage to get full power.
Title: Re: JMP 2100 not making power
Post by: ilyaa on March 30, 2014, 05:18:21 AM
k so i got new el34s, popped them in, but everything is *not* better.

1) even with the bias voltage as high as it will go (about ~30VDC), the tubes are only pulling 30 mA or so idle current. if i swap the old tubes back in, they pull way more. ??. the amp chassis says 6550 on it, and ive read that these amps were kind of made for those tubes. is there something i need to change in the bias circuit for el 34s to work? if that's the case, though, can someone explain why the old tubes (also el34s) could pull so much more current? but still make low power for the amp....?!

2) measuring at the plates of the PI, i only get about 40V p-p. that is definitely less than twice the bias voltage. ??

3) ive got about 400 VDC on the power tube plates and pin 4s. right heater voltage, too. no other suspicious stuff that i can see.....

Title: Re: JMP 2100 not making power
Post by: DrGonz78 on March 30, 2014, 06:14:48 AM
I have only read about such issues with these types of amps so take my advice with a grain of salt. However, there are versions of these amps where there were 6550 or KT88's for the output tubes. The bias resistors for EL34's would be the 220k ones that you see in your schematic. However, if you had 6550/KT88 type then those resistors would be 150k. So it would be wise to look at your amp and locate those resistors to see exactly what we are dealing with here. If those are 150k resistors then yeah the tubes won't bias correctly with EL34's. Good luck. Look at the schematic attached...
Title: Re: JMP 2100 not making power
Post by: DrGonz78 on March 30, 2014, 06:21:15 AM
Please note that I am wrong calling those 220k resistors "bias resistors" and I believe actually they are called grid leak resistors. However, the bias circuit could be wrong in general terms and you need to start by measuring/confirming all the values in relation between both schematics on this thread to your amp.

Edit: The grid leak resistors are not the main problem to bias the outputs as the one single 150k or single 220k resistor that is in the bias circuit. Not sure also if the Bias trim pot is different but I don't think it is...?
Title: Re: JMP 2100 not making power
Post by: Roly on March 30, 2014, 07:38:57 AM
Quote from: ilyaa2) measuring at the plates of the PI, i only get about 40V p-p. that is definitely less than twice the bias voltage. ??

Ah, you should have said that sooner.  That doesn't look right to me, a differential amp PI should be capable of a lot more signal swing.

A quick LTSpice suggests that it should just be starting to limit with 4 to 5Vp-p in, at which point you should be getting around 180Vpeak-peak on each anode.

Now the question is, is this the PI's fault, or is it getting insufficient drive from back up the signal chain?

Check that you have ample HT on the feed point for that stage, and the next obvious thing to try would be a known good ECC83/12AX7, however measure the voltages around the PI and see if they look sensible - if they do then the problem is likely to be earlier in the signal chain.

With the EQ all set for 50% you should be able to inject a signal into the wiper of the treble control and be able to drive the OP stage to full wick (it will take a few volts, and the OP stage should clip long before the PI).

Check that the anode voltages in each preamp triode section are reasonable (i.e. a bit over half the local HT).

If you still aren't getting any joy I'd try injecting a signal in the front end, say around 100mV, and check that you are getting something like x30 voltage gain from each triode section (except the cathode follower feeding the tonestack of course, that will be close to unity).
Title: Re: JMP 2100 not making power
Post by: g1 on March 30, 2014, 01:54:28 PM
  The bias range will be greatly affected by the resistor Dr.Gonz mentioned, at the cathode end of the bias diode.  It and the 2 resistors at the power tube grids should be changed for the type of output tubes:  150K in all 3 positions for KT88, 220K for EL34.
  As far as the low volume, a good place to start is DC voltages for pins 1,3,6,&8 of V1,V2, and V3.
Title: Re: JMP 2100 not making power
Post by: ilyaa on March 30, 2014, 02:20:45 PM
yeah i think its set up for 6550, according to the bias network.
but still, can anyone explain why two different sets of el34s would be allowing such drastically different cathode currents?

DC voltages:

V1:
pin 1: 100V
pin 3: 0.7 V
pin 6: 140V
pin 8: 1.4 V

V2:
pin 1: 120V
pin 3: 0 V (!)
pin 6: 220V
pin 8: 129 V(!!)

V3:

pin 1: 200V
pin 3: 34 V
pin 6: 200V
pin 8: 34V

injecting a signal straight into the treble control i can still only get about 60V p-p off the PI, though....doesnt this bypass V1 and V2, anyway?
Title: Re: JMP 2100 not making power
Post by: g1 on March 30, 2014, 08:55:28 PM
  The difference in bias of the power tubes could just be variation in the tubes, or the old ones could be faulty.  Once you change the bias circuit to the proper resistors for EL34 you can get a better idea of how much difference there is (how much difference in grid voltage to get the same cathode current).
  Yes, injecting your signal at the treble pot wiper bypasses V1&2.
So what is happening when you hit 60Vp-p?  The waveform is clipping?  Do you still have lots of room to increase the output of the generator?
Try it with the power tubes removed and see if you get more than 60p-p.

P.S. You said you were only getting 40Vp-p, but now you are getting 60Vp-p, which is double the bias voltage if it is at -30V.  Is this the case?
And when the PI is putting out 60Vp-p, what voltage do you get at the load?
Also, V2 pin3 appears to be a measurement error, recheck.
Title: Re: JMP 2100 not making power
Post by: ilyaa on March 31, 2014, 05:11:38 PM
okay:

putting in a 1khz 100 mV sine wave into the input (with no power tubes installed) -

stage one (measured at the anode): 2 V p-p
stage two (measured at cathode): 45 V p-p (with volume cranked)
stage three (measured at anodes): 100 V p-p (starts to clip)

should stage two be at unity? overall this seems like okay gain, right? about x1000. thats a big swing off the PI, right?

putting a 1khz 1 V sine wave into the treble control (no power tubes) -

stage three (measured at anode) 100 V p-p (starts to clip) - about the same.

putting in a 1khz 100 mV sine wave into the input (with power tubes installed) -

stage one (measured at the anode): 1 V p-p
stage two (measured at cathode): 30 V p-p (with volume cranked)
stage three (measured at anodes): 75 V p-p (starts to clip)
voltage at the 4 ohm load: 35 V p-p


this is with the not properly biased el34s in. i have some 6550s i will put in later on (a friend is bringing them). seems like the amp might be in okay shape, overall though, right? maybe just needed the preamp tubes moved around (which i did).

also i rechecked and yes, i get about 0.7V at V2 pin 3. but all the other DC measurements are the same as i posted previously.

strange detail: the presence knob has been re-wired on this amp. instead of being attached to the grid of the PI, it is wired as a rheostat between the two 22nF caps on the PI anodes. if this knob is all the way down, i cant get nearly as much unclipped power out of the PI as I can if i crank it up. what is this modification?
Title: Re: JMP 2100 not making power
Post by: Roly on April 01, 2014, 09:59:08 AM
Quote from: ilyaait is wired as a rheostat between the two 22nF caps on the PI anodes. if this knob is all the way down, i cant get nearly as much unclipped power out of the PI as I can if i crank it up.

If I understand you correctly, a variable resistance simply wired between the anodes of the PI (no cap in series) will only shunt/short the drive voltage to the OP stage, so what you observe is what I'd expect.

Quote from: ilyaawhat is this modification?

um ... clueless?  It seems like a mighty curious mod to me.

If you look at the circuit the NFB is applied from the OPT secondary via the 47k(100k) resistor to the cathode circuit of the PI, and the cap on the pot wiper acts as a top cut on the NFB.  If you top cut NFB you end up with top boost, but if you don't have a cap at all you don't have a frequency selective component and the knob loses any claim to being a tone control.  Unless you have a compelling reason to leave it I'd restore it as original because this will certainly limit available OP power by limiting the drive into the OP grids.

Quote from: ilyaavoltage at the 4 ohm load: 35 V p-p

P=E2/R

35pk-pk = 17.5Vpk = 17.5/1.414 = 12.4Vrms

(12.42)/4 = 38.44 watts, which is not unreasonable, but I'd expect more like 50-60 watts from a pair of EL34's in fixed bias.

Are you sure this is supposed to have a 4 ohm load and not 8 ohms and/or it is on the right OPT tapping?
Title: Re: JMP 2100 not making power
Post by: ilyaa on April 01, 2014, 11:33:59 AM
yes - the mod on the PI output has a direct impact on the output voltage. i think i will restore it - is it some kind of strange effort to put a master volume on the final stage?

it has an output selector. it was set to 8 ohms originally but i only made a 4 ohm dummy load (what a dummy) and so ive switched it to be on 4 ohms. should i trust the selector? if not, can i measure resistance on output transformer leads to verify?

Quote(except the cathode follower feeding the tonestack of course, that will be close to unity).

should that stage have been giving me unity gain?
Title: Re: JMP 2100 not making power
Post by: g1 on April 01, 2014, 05:14:53 PM
  The cathode follower is not a gain stage, it is for driving lower impedances like the tone stack so it is fine.
Which side of the 22nF caps was the pot on?  At the PI plates or the power tube grids?
  It's acting as a master, but will only go to zero if the PI is perfectly balanced.  As that is not likely, you will still get some sound with it turned down all the way.
What did they use to replace the presence pot, just a 5K resistor with no cap?  That would be like having the presence turned all the way down all the time.
  Check with the owner if he uses that "master", if he does, I would recommend moving it somewhere and putting the presence circuit back to stock.
Title: Re: JMP 2100 not making power
Post by: Roly on April 02, 2014, 05:52:48 AM
Quote from: ilyaashould i trust the selector? if not, can i measure resistance on output transformer leads to verify?

a) Yes.

b) mmmm doubtful.  The OPT is like a gearbox, and the speaker is the load on that gearbox, the "torque" of the load being "reflected" by the ratio back to the OP valve anodes.  The resistance you measure on the OPT secondary is just the resistance of the winding.  This can be useful for sorting out unmarked windings because a higher impedance winding will have more turns, more wire, and thus more resistance, but the resistance doesn't tell you anything about the ratio of turns between the primary and secondary, only respective steady-state mid-band AC voltage measurements between primary and secondary can do that.

The turns ratio of a transformer is the same as the voltage ratio.

The impedance ratio is the square of the turns or voltage ratio.

Thus a transformer with a turns ratio of 4:1 will also have a voltage ratio of 4:1 but an impedance ratio of 4 x 4 = 16:1.

To a first order approximation; a transformer doesn't have an inherent impedance, what you see on one side is the impedance connected to the other side "transformed" by the transformer impedance ratio.

Typical conditions for 6CA7/EL34 are;
54W out, 3.5k plate-plate, typical load ZL = 8 ohms, meaning an impedance ratio Zp/Zs of 437.5:1, from a turns/voltage ratio Vp/Vs of 20.9:1.


Here's some stuff to paste in your hat about Voltage Followers

In voltage follower stages, valve/FET/transistor/op-amp, the input is to the grid/gate/base/+input and the output is from the cathode/source/emitter/output, the anode/drain/collector generally being connected directly to the supply.

The follower has unity gain (strictly slightly less, ~ x0.999).

The input impedance is high, typically megohms.

The output impedance is low, typically ohms.

The function of a follower is to act as a buffer to isolate a high impedance source from a low impedance load.  You will find voltage followers driving tonestacks, as here, and in Line Out stages to drive long high capacitance cables without suffering treble loss.
Title: Re: JMP 2100 not making power
Post by: ilyaa on April 02, 2014, 11:46:03 AM
QuoteTypical conditions for 6CA7/EL34 are;
54W out, 3.5k plate-plate, typical load ZL = 8 ohms, meaning an impedance ratio Zp/Zs of 437.5:1, from a turns/voltage ratio Vp/Vs of 20.9:1.

where do you get the 3.5k?

Quoterespective steady-state mid-band AC voltage measurements between primary and secondary can do that.

whats that?

as for voltage followers, i do understand the role/function of a buffer. what i dont understand, is why my measurements seem to give me gain across the cathode follower.  1 V p-p in (from anode of the first triode) and 30 V p-p out (off cathode of the cathode follower)
Title: Re: JMP 2100 not making power
Post by: Roly on April 02, 2014, 01:46:46 PM
Quote from: ilyaawhere do you get the 3.5k?

It is a typical figure for 6CA7/EL34's derived by drawing a load line on the anode characteristic curve of the valve in question, between the maximum current point on the Y-axis to the maximum voltage point on the X-axis, the slope of which represents a particular load value.  It's one of the many sets of conditions that the valve manufacturer helpfully calculates for us, and which the vast majority of users follow.

This is a typical load line (which is for a pair of EL84's);

(https://lh3.googleusercontent.com/-QJI_aji5C3M/T1DnQ03fRvI/AAAAAAAAADw/SmrjxGwOsDk/s720/Fellow%2520PA%2520Loadline%2520Class%2520A.JPG)

The red line is the load line running from a supply of 288V x2 = 576V (actually drawn 285V and 550V) to a maximum cathode current of 90mA, the slope being for 3.2k (per side) which is a quarter of the final 12.6kplate-to-plate, the required bias being -10.5V.

The green lines show the idle voltage and current conditions and just intersect with the maximum anode dissipation curve of 12 watts (which you normally have to plot yourself).

This is a load line for a classic 12AX7 triode stage;

(http://stevepedwards.com/wp/wp-content/uploads/2014/02/020214_0244_LaneyKlipp113.png)

Here the load line is in blue and runs between the supply voltage of 300V at (A, valve cut off) and the maximum valve current of 3mA at (B, valve saturated), the slope representing an anode load of 100k.  The Q or idle point is a little above half the HT supply voltage.  Illustrated is a 1V change in grid voltage from -1V to -2V producing a change of 0.6mA in the valve current, resulting in a 60V change in the voltage across the 100k load.

Note the different general shape of pentode vs. triode curves.


Quote from: Rolysteady-state mid-band AC voltage measurements between primary and secondary

Steady-state; a continuous sine wave of unchanging amplitude.

mid-band; away from the natural high and low frequency roll off of the transformer, typically 1kHz or 440Hz.

AC voltage measurement; with suitable AC voltmeter or CRO.

between primary and secondary; the ratio of the voltages measured on each side of the transformer, Vprimary/Vsecondary = voltage ratio = turns ratio = the square root of the impedance ratio, i.e. square the voltage ratio to get the impedance ratio.

E.g.
If a transformer is known to have a voltage ratio of 21:1 then the impedance ratio will be;

21 * 21 = 441:1

If we connect an 8 ohm load to the secondary then the impedance across the whole of the primary (plate-to-plate) will be;

8 * 441 = 3528 or 3.528kp-p


Quote from: ilyaawhat i dont understand, is why my measurements seem to give me gain across the cathode follower.  1 V p-p in (from anode of the first triode) and 30 V p-p out (off cathode of the cathode follower)

Well since that is actually impossible there must be another reason.

The stage before should provide a gain of about x30 between its grid and its anode (which is tied to the grid of the follower), so it's Sydney to a brick that what you think is the preceding stage anode is actually its grid that you are looking at.  {Or perhaps you are the first person to break Ohm's Law ...

um ...   ???

no, you have confused which pin you were probing.}


ed typo
Title: Re: JMP 2100 not making power
Post by: J M Fahey on April 02, 2014, 08:23:02 PM
To make sure, read first AC signal at the cathode follower's own cathode, then measure signal at its own grid (not the other triode's plate).
Use a high impedance, AC coupled meter, such as a scope or a good multimeter .
Title: Re: JMP 2100 not making power
Post by: g1 on April 03, 2014, 12:16:34 PM
Agree with both above.  Looking back at how you called out the stages, you have to look at V2 as two stages, not one.

It appears you have sufficient drive at the power tubes and reasonable output power.  New power tubes properly biased will probably get you a bit more power out which should be in the ballpark of normal for this model.

One thing to note is that power tube grids can't be driven beyond the bias voltage, they will just clip the signal.  So for example if your bias is -30V, anything beyond 60Vp-p at the power tube grids will get clipped, even if the PI is capable of more output level (like you saw with power tubes removed).
Title: Re: JMP 2100 not making power
Post by: ilyaa on April 03, 2014, 01:20:38 PM
oh so in the first half of V2, that anode goes to the grid in the second half of V2? is that right? so there is voltage gain there to feed the cathode follower?

didnt realize that tube was two stages.
Title: Re: JMP 2100 not making power
Post by: J M Fahey on April 03, 2014, 02:16:43 PM
Yes, first triode supplies voltage gain, second buffers that output.
ThereĀ“s your 30X gain which sounds very reasonable.
Title: Re: JMP 2100 not making power
Post by: ilyaa on April 07, 2014, 02:49:03 PM
alright, we're good to go:

there had been a mysterious knob on the back panel of the chassis that i was just ignoring - that was the presence knob, rewired just like the schematic.

i removed the weird master volume mod and rewired the normal presence knob back to stock (left that extra knob on the back disconnected).

got some 6550s and put them in and biased them (they biased right no problem). one of the screen grid resistors blew up while i was burning them in, but i replaced it and things seem to be fine. thinking it was just old and ready to go.....

everything else seemed to be fine. amp sounds great! loud and really detailed sounding, with all kinds of surprising harmonics that come through with dynamic playing. especially once you pass a sweet spot on the volume and things start to break up. it was putting out about 13-14 Vrms across 4 ohms (so almost 50W) - i think we're all good!

thanks, guys - another amp back to life. and i learned some stuff!

Title: Re: JMP 2100 not making power
Post by: Roly on April 08, 2014, 10:20:28 AM
Not thrilled about the screen resistor burning up - that is not a good sign.

This time you encounter a different aspect of guitar amp servicing, The Mad Modder.  Many times we stand there looking at some modded amp chassis, scratching our heads and wondering "What WERE they thinking?".  Sometimes you can get a clue as to what they had in mind, wrong though it may have been, but other times it just plumb evades you.  Generally the safest thing is to go back to original.

An oldie, but an extreme goodie;
https://www.youtube.com/watch?feature=player_embedded&v=yDJibf8uZnQ (https://www.youtube.com/watch?feature=player_embedded&v=yDJibf8uZnQ)
Title: Re: JMP 2100 not making power
Post by: ilyaa on April 08, 2014, 10:52:59 AM
i think it may have burned up because i flipped the amp on without waiting for it to warm up when i was burning the tubes in - could the surging power/current have done the resistor in?
Title: Re: JMP 2100 not making power
Post by: Roly on April 09, 2014, 04:00:40 AM
What "surging current" @ilyaa?  If you plan on making a hobby/career out of repairing amplifiers you will have to get coldly calculating and seriously rational about your circuit analysis.


Let's go back to some basics.

How does conduction occur in a valve?  Electrons, right?

And where do these electrons come from?  The heated cathode, right?

How many electrons come off a cathode that is cold?  None, right?

So how could any current, surge or otherwise, occur before the cathode has heated up?

The simple fact is that it can't, and that you are another victim of the widespread "standby switch mythology".

Since I'm inherently lazy, you will find one of my demolition jobs on standby switching mythology here;

http://www.guitargear.net.au/discussion/index.php?topic=40125.msg442774#msg442774 (http://www.guitargear.net.au/discussion/index.php?topic=40125.msg442774#msg442774)

If you give it a bit of thought you might conclude that standby switching actually causes "surges"!

In a typical guitar amp the standby switch opens the HT rail somewhere, and the result is that all the bypass electros downstream from there are discharged through bleeders or hot preamp valves.  Now when the standby switch is closed again all this capacitance has to charge up, and the result is a surge current through the power supply, which in general isn't good for anything (least of all the standby switch itself).

Cold valve with HT applied from a solid-state rectifier.  No heat, no electrons, no conduction.  But as the cathode slowly heats up so the electron emission slowly starts and the valve slowly comes into conduction.

Pre-heated valve, up to temperature with the standby switch open (as widely advised).  Lots of electrons, lots of potential conduction.  So then you close the standby switch and suddenly apply peak HT volts from the supply to the output valves, and WALLOP! a surge of current goes through the output valves until the power supply is discharged back down to its loaded working voltage.

Now which situation do you think might be more damaging for the valves?

Well it's a trick question because valves are well able to cope with either situation.

If you examine the history of guitar amps you discover that many of the early circuit arrangements were lifted from Ham radio AM modulators, and that these often had standby switches to idle them when the Ham station was in receive mode.

This obviously doesn't apply to guitar amps, but the standby switch seems to have made the transition in spite of that, and its only apparent function being to disable the amp between sets during breaks without disturbing the amp controls.

For a lot of people (who would like to impress with their ...ahem... grasp of the technical) this alone wasn't enough, so a whole bunch of Old Wives Tales were born to make the standby switch more "techie" than it actually is.

So when you hear these myths, just remember that the valve guitar amp is just about the only valve appliance that ever had a standby switch, and that you will look in vain at millions of mantle radios, radiograms, tape recorders, b/w TV's, &c&c, and never see a standby switch.
Title: Re: JMP 2100 not making power
Post by: ilyaa on April 10, 2014, 03:03:34 AM
first of all, that mod video is insane. i can hardly believe its real.....

what might have burned the screen resistor? normally is that a sign of a bad tube, pulling too much current?

Roly, what you say about standby switching makes a lot of sense - I never thought about it that way before. I read some of the other links and see that the standby myth has plenty of debunkers.

a related question: what about the standby switch and cap DIScharging? with that sound city ive been working on, i see that if i turn the standby switch off (open) before i switch the amp off the caps drain quite slowly. BUT if i switch the amp off and leave standby on (closed) they drain much more quickly. i guess thats because with the standby switch closed, the caps can drain through all kinds of stuff throughout the amp, but with it open, they can only drain through those rather large bleeder resistors. ive also noticed though, that if i leave it closed when i turn the amp off and let the caps drain, they start to charge back up if i open it again (even if they have drained to 0V....)....whats charging them, in this case??
Title: Re: JMP 2100 not making power
Post by: Roly on April 10, 2014, 01:27:48 PM
There are a limited number of ways of being right or sane, but there are an unlimited number of ways of being wrong or insane.  If you go on fixing amps the chances are one day you will encounter something that tops this.

I once had somebody bring me an amp kit they had built.  Generally techs try and avoid new builds like this because any and everything could be wrong with something that has never worked.  At least with an amp that has been working it's generally only one thing wrong.

Well this lady didn't know how to solder, so she had used rubber cement for all the joints.  After I got over my initial shock and took a close look she had followed the instructions carefully and actually made quite a neat job of it.  In the end I took the chassis outside, fired up a high powered soldering iron, and simply soldered through all the glue.

"Stinks like hell, the neighbours complain,
I don't give a hoot what you say,
I don't give a hoot what you say"

http://www.youtube.com/watch?v=eIVOt4fbs0U (http://www.youtube.com/watch?v=eIVOt4fbs0U)

With all the joints properly soldered it worked just fine first time.


Fried screen resistors aren't all that common except around 6CA7/EL34's were the screens are prone to collapse and short to the cathode.  6550's aren't terribly common here and I've never had any particular problem with them, so I can't say.  The only guess I can make for a self-curing fault like this is perhaps the anode connection in the socket was a bit dodgy and went open, but that's very much a shot in the dark.  A resistor may go open by itself, but it won't cook and burn up by itself.  Maybe one of the other guys with more experience of 6550's will comment.


Yes, your observation about cap draining time is correct; when the hot valves are left connected they will also drag the supply down.

And now you make the astute observation that caps can regain charge after fully discharging.  There is some debate about the exact mechanism but it is likely to be related to dielectric relaxation.

Basically when you discharge a capacitor to zero volts you don't actually remove all the stored charge, some remains trapped within the insulating layer.  If the cap is then un-shorted this charge will appear as a small voltage gain as it leaks out of the dielectric to the conductive plates on its surface.  The degree to which this happens depends of the dielectric material and where accuracy of stored voltage is important, such as in instruments, then high quality caps with specially formulated dielectrics that exhibit minimum "dielectric relaxation".  Electrolytic caps are built to store the most charge in the least space so normally have the greatest dielectric relaxation effect.

This is why I leave my discharge lamp clipped on when I'm working on an amp that has just been powered up.  The low resistance of the lamp will just let the residual charge leak away, and if I should happen to forget to remove it it will light up to remind me with no harm done.

Discussion here;
http://blog.shuningbian.net/2005/04/self-charging-behaviour-of-electrolyte.php (http://blog.shuningbian.net/2005/04/self-charging-behaviour-of-electrolyte.php)