New Amplifier Design - "Murder One" - Submini Pentode, Low Voltage

[frequencycentral]

Sounds a little off. Something's wrong there.....cold joint again? Also check for shorts in the multiplier - it should be putting out a few more volts than that.

[bluebunny]

Thanks Rick, I shall take an even-closer look at my handiwork.  I was post-gig tired yesterday, so anything could have happened...     BTW, how many "few more volts" should I be expecting?  I'm assuming after each diode I should see another 12V (11.7V!) less a diode drop?

[bluebunny]

DMM has done its bit.  Here's how the charge pump is looking:

After D15, 15.7V.
D2 = 20.0V
D4 = 20.2V !
D6 = 36.4V
. . .

OK, even I can see a break in the pattern here!    Looks like something went awry around D3/D4 in the chain?  I'm assuming that since B+ is not zero, this is more likely to be short- than open-circuit?  I've been staring at this area through a magnifying glass for many, many minutes and can't see anything obviously wrong.  Does a fried 1N4148 go short-circuit?  And I thought I'd been deadly efficient with my soldering.  Perhaps just deadly...    Actually, the diode/continuity test seems to show all the diodes with a small forward resistance and open backwards, so perhaps they're all fine and dandy.

Am I getting warm, or is that just the 6111?  Will continue to peer and probe in the meantime...

[bluebunny]

Ooh, my bad!  DMM was having a little brain-fart of its own.  (I know how it feels.)  The middle segment of digit two was out to lunch.  So 20.2V was actually 28.2V!

Anyways, that means the voltage is climbing the ladder alternately ~4.4V and ~3.8V per rung.  And I'm none the wiser...  

EDIT: btw, is the 7660 supposed to be getting rather warm?  After the 6111, it's the warmest thing on the board.

[frequencycentral]

A warm 7660 is indicative of an error somewhere. Really double check there is no hairline shorts in the multiplier, particularly around the 7660 pins 2 and 3. You may not even see it, it'll be that fine. If ever I'm in doubt I gouge away with the thin edge of tweezers until I'm 100%.

[sadcakes]

Hi,
I can't offer any help, but I have also just finished my build using the V7 PCB and I am experiencing the exact issue that Bluebunny is.
With both gain and Volume up full there is a tremolo style oscillation.
Reducing either control to about 8 removes this.
I also am getting 11.68v across pins 1 and 8 of the 7660S.
Very interested to hear if you find the source of your issue.   
Sean

[sadcakes]

Hi Rick, I've removed the (hot) 7660 and checked for shorts. I am getting a reading between pins 2 and 3, when all the others are open.
Is this an indication of a faulty IC?
Thanks
Sean

[bluebunny]

Thanks for the advice again Rick.  I don't think there's a short between 2 and 3 'cause they show different signals on my little oscilloscope.  But I shall have a good gouge anywhere there are close tracks.  But will have to wait until tomorrow - will let you know how it pans out, Sean.  I'm past tired and being awake is probably a good idea for continuing with this one!  Up early tomorrow morning (like always) so kip is sadly #1 priority right now.  Even my daughter has just come in and said "aren't you supposed to be in bed?".  Bloody cheek...  

[Blackface007]

Sorry, I have deleted my posting, to avoid any confusion.

[bluebunny]

OK, been over it with a fine-tooth magnifying glass and gouge!  No unexpected shorts.  7660 still getting warm.  One thing I did notice: the "tremolo" effect at high gain seems to be a sub-sonic pulsing.  Having the benefit of daylight (working at home today), I can see the speaker is bobbing up and down gently but inaudibly.

Anyway, I was wondering about duff or rogue parts - particularly since you indicated warm 7660s aren't a good sign.  I'll try swapping it out and see what happens.

I'm determined to sort this out.  It's a great design and I have an amusing "re-purposed" enclosure sitting waiting for it...

[bluebunny]

New 7660 made no difference.  This one from HK, the first was from Rapid.  Speculated offline with sadcakes that the 7660 might be hot because of its proximity to that 18R voltage-drop resistor?  Contrary to what I said earlier, this is the hottest point on the board - close to boiling!

[sadcakes]

Ok, I definitely need some help now!
I've just completed build version 2 using the second of Rick's v7 boards that I ordered and some new components.
I have checked the build as thoroughly as I did the first at every point along the assembly.
Despite this, this version is experiencing the exact same effects as my first build. With both Gain and Volume on full there is a tremolo like oscillation.
Can anybody who has successfully built  the V7 using the PCBs perhaps assist?
Could it be my off board wiring??
thanks a lot
Sean

[sadcakes]

So I managed to fix this myself..

Using Rick's advice to Blackface007 I swapped out R1 and R2 for 100k values and also replaced R12 with a 470k.

This has removed the 'motor-boating' oscillation I was experiencing at full volume even though I am using a 6111 tube as per the schematic.

Bluebunny - you may want to give this a try!

The only issue I have now is excessive background noise/hiss at full volume but I'm hoping this will be reduced once I have it in an enclosure...

Oh and if anybody can offer an idiots guide how to wire up the Pre-out switch (currently I have the orange pads jumpered) that would be much appreciated! For some reason the illustrated pic makes no sense to me!

[frequencycentral]

Here ya go!

_______________________________________________________________________________________________________________

Basic build:



________________________________________________________________________________________________________________

Added switch to select preamp or power amp at output socket:



What's happening in the above layout is that the preamp out is taken from lug 2 of the volume control. One half of the DPDT selects either the preamp or the power amp to the output socket. The other half of the DPDT cuts HV to the pentode and the output transformer when the DPDT is in preamp mode - as it's bad for the pentode and transformer to see high voltage without a speaker attached. I'd really stress that caution needs to be taken with this switch, as you don't want to send the power amp output into anything other than a speaker. It's a cool mod though - in preamp mode you've got a massive warm tube overdrive 'pedal' type effect.

_______________________________________________________________________________________________________________

So....there is a third way to wire up the Murder One, which is to include separate sockets for preamp out and speaker out. If you do this you'd want to use a switched socket for the speaker out and include a dummy load across it, so that the power amp and transformer are seeing a load even when there is no speaker plugged in. If anyone wants to build it this way I'll work up a wiring diagram.

[sadcakes]

Rick, you are a star!
That's perfect, thank you.
Sean

[bluebunny]

Thanks Sean, I'll give those resistor mods a try.  I'm WFH today - snowed in (that's what I told the office!) - so I may have a go during my "lunch break". 

[bluebunny]

Wey hey!  No more tremolo!  Thanks for the heads up, Sean (and Rick's advice to blackface007, which I'd clearly missed...).  Now I just need to get to the bottom of the measly ~62V at the end of the charge pump, cos, how can I put this? - it's a bit quiet!

[iccaros]

Wey hey!  No more tremolo!  Thanks for the heads up, Sean (and Rick's advice to blackface007, which I'd clearly missed...).  Now I just need to get to the bottom of the measly ~62V at the end of the charge pump, cos, how can I put this? - it's a bit quiet!

which diodes are your using? some drop more than others, can you post your voltages for each drop measured off C2, C4, C6 ect...

charge pumps work by packing electrons in to the capacitors, by oscillating, the capacitor dumps across the diode, when packing more electrons across the diodes you get more voltage. With this design its a latter where you get a voltage with no load, once there is a load there are not enough electrons left to = said voltage and you see the voltage drop.
you could have a power supply that is not supplying enough current causing this supply to SAG.
Also this tube puts out 65mw. I have built amps using 12k5 which have the same power output and they are quite, this is not a jam with the band amp, but a tone amp. add an inefficient speaker and its worse.  I use a Red Coat Wizard which is a 103db @ 1watt. In a similar amp I get 70db at one foot, according to my meter. 70db is quite, but loud enough to record or mic.

if you built off the board design, you could try some things, different wall wart, or use 13v supply. this is over the chip rating, and with Rick's voltage drop raises the heater voltage on the power tube. So is not a great option .

Schottky diodes will drop less voltage, giving you a slightly higher voltage at the end. A BAT41 is rated at 100 peek inverse voltage and 100ma. so it should work with no issues. the 1N5711 can be used also up to that last two diodes in the chain. Then your getting to close to its 70v max inverse voltage. 

[bluebunny]

Thanks for the additional comments, iccaros.  I've got a string of 1N4148s, as per Rick's design.  I'll power up again this evening when I'm home and check the voltage at each step.  The PSU should be fine: it's a 1A 7812-based job - hopefully plenty for this and one or two more of Rick's tube pedals...     The speaker is culled from a TC-Helicon VSM300XT, so I'm hoping it's not too inefficient (though I readily admit something more "guitar" might be more appropriate).  Will keep in mind the diode sub suggestion, though I do *hate* de-/re-soldering!

[bluebunny]

OK, here's the story down that charge pump:

C2 = 20.3V
C4 = 28.8V
C6 = 37.3V
C8 = 45.8V
C10 = 54.3V
C12 = 62.8V

By my calculations, and according to the 7660S datasheet, it should increase by my starting 12V less two 1N4148 drops each step?  Assuming 1V V_F per diode (again according to the datasheet) I should get 72V at the end?  Where'd my other 9.2V go?