advertisement


Mystery amp schematic - assistance/comments/guesses/insights welcomed

OK, I'm back. I've measured (or re-measured) everything. Not everything makes sense to me.

First off, all measurements relative to PCB gnd (the green wire "centre tap"). Scope input coupling mode set to AC.
1: 0
2: 1mV, 50Hz
3: 1mV, 50Hz
4: 0
5: 1mV, 50Hz

Virtual gnd is at 0V relative to PCB gnd, (so no measurable ripple on the 100uF)
re-measuring relative to virtual gnd just to be sure - everything's the same.
1: 0
2: 1mV, 50Hz
3: 1mV, 50Hz
4: 0
5: 1mV, 50Hz

So the purple and red leads running from the input sockets are at 0V, they don't appear to be picking anything up, despite their proximity to the AC supply to the PCB.

Measuring across things
1-2: 1mV, 50Hz
2-3: 18mV, 50Hz (much bigger than I thought it could be)
3-4: 150mV, 50Hz (WTAF!)
3-5: 70mV, 50Hz (not so surprising after the previous one, but certainly unexplained)

I am open to suggestions, including why-and-how my scope is lying to me!
 
I think your scope is just adding noise pickup into the loop. Don't forget that with no plug in the input that 1 and 4 are connected which forms a loop 1-2-3-4-1. Do you have a bare plug with nothing connected to it? Try that in the input (to break the loop) and see if it quietens it down. I'm pretty certain it's a loop problem because shorting across 1 and 2 as you did above reduces the mains pick up by removing the impedance of the 2n2 from the loop.
 
OK. Quick summary of the (complicated) input jack wiring/switching.

Green wire is PCB earth, red and purple are on the schematic. All wire are unshielded, single core.

I *do* have a bare plug (actually a handy phono to 1/4 jack adapter...)

With nothing in either socket, both red and purple are connected to green.

With a jack in "normal" (on the front panel), the tip of the 1/4" jack goes to red, the shaft of the jack plug goes to green and purple.

With a jack in "bright" (on the front panel) the tip goes to red and purple (a loop, but not earthed this time); the shaft goes to green.

All this was checked with continuity mode on my multimeter.

With the bare plug in "normal", I get 6mV, 50Hz at 5 (op amp input)

With the bare plug in "bright" I get 18mV, 50Hz at 5. This is still a loop, but it's now not even grounded!
 
Really not sure where to go from here TBH. Anyway, there's a lot of high impedances on the input to oa2 and these are earthed via 100k on the non-inverting input pin. Plugged into bright, you have effectively isolated the loop, but the whole loop is now floating, connected via 1n and 100k to earth. (As an aside, are you sure it is 1n as that with 100k is a corner frequency of 1.6kHz? I notice a 0.01 or 10n in the circuit nearby which would make more sense). Plugged into normal, red and 68k is floating, connected to earth via 2k2 + 2n2 in parallel with the DC blocking cap (your 1n) + 100k, both are high impedances but these form a loop through the 100uf capacitor.
I really think you need to reduce the loop area of red green purple (even though plugging into bright seems to indicate it's not this) as it won't do any harm. Might also help to move the low voltage ac wires away from the jack socket. By the way, do the jack earth's connect to the front panel or casework?
 
I think we're looking at "the way it is". I've seen another one on eBay (just the PCB and front panel) and as far as I can tell from slightly dodgy/blurred photos it's wired the same.

I have also simulated the normal/bright circuit (and wiring) and it does indeed work. "Normal" is actually a big/wide peak in the midrange, and "bright" is flat to the right of the peak.

EDIT; the jack socket's "earth" is only connected to PCB "earth".

So we're trying to fix the design, not repair a damaged/broken item.
 
Last edited:
Right; an expanded statement and simulation of "how things are" (click small image to actually see the traces on the large image)

Badger input simulations by plybench, on Flickr

The "bright" input takes the signal to both red and purple wires; this is simulation (3), red trace, showing bass dropoff below 1KhZ

The "normal" input takes the signal to the red wire, leaving purple at gnd; this is simulation (1), green trace. This dumps some treble to ground via the 2200pF capacitor, and gives a frequency response with a midrange peak at around 400Hz.

This switching of tone via the selection of an input socket appears to be a "homage" to Fender's guitar amplifiers.

I have also simulated "purple only" (which is not a wired possibility). Simulation (2), blue trace. This brings the signal through both capacitors, giving a stronger low frequency drop below 1300Hz.

I have done a second simulation of "normal", changing the 2200pF capacitor to other values (as labelled).
 
I propose (as a trial) desoldering all the leads (earth and two signal) from the board. I think that should make most of the hum and trash go away.

Or would I then need to actively ground (or something) the inputs to stop them "floating"?

If this does bring blessed silence, I then propose ONLY wiring the "normal" input socket with a shielded lead, taking this wire to the PCB input, and then taking the other channel to ground. This is intended to give me a "normal" only amplifier that is quiet.

I could also tune the frequency response a little per my second simulation, but in any case, the amplifier has extensive tone manipulation controls over-and-above the twin socket stuff.
 
Last edited:
OK. I have experimentally desoldered the input leads and earth from the PCB, and (on the back of the PCB) put minimum length jumpers from PCB-gnd to the 2 inputs.

Result: about 2mV 50Hz on the input of OA2, and 25mV at the output. :(:mad:

i.e. just as bad as having the real leads present.

This hum is startin' to look permanent.
 
Last edited:
Right (Rolls up sleeves).

Having tried PCB-gnd-ing the lead-in points (node 1 and 4), to no avail, I went radical. For the sake of a bit of wire and some solder, I wired node 3 to PCB-gnd.

At last! :):) OA2 positive input is now a 0V (AC), and OA2 output is down to 4mV, possibly from a tiny amount of ripple on the virtual gnd.

I'm not 100% sure that this tells me - it's pretty straight forward that connecting the input (node 3) directly to PCB-gnd should stop hum.

Although it's equally straight forward that connecting nodes 1 and 4 to PCB-gnd should also stop hum, and yet it didn't.:(

Not sure what to try next. :confused:
 
Last desperate flail around:

Removed the grounding wire from node (3), and moved it to node (2). Now nodes 1,4,2 are connected to PCB-gnd.

This results in a tiny signal (just under 1mV, no measurable frequency) on (5) the positive input of OA2, and 5-6 mV, 50 Hz on the output of OA2.

It appears the entire component train is an effective absorber of AC hum.:eek:
 
It tells you (all other things being equal) that if you move the earth from #1 to #2, the noise is somewhat reduced, and moving it again from #2 to #3, the noise is reduced slightly further on the output of oa2 because you are reducing the impedance to the non inverting input to oa2. I reckon it must be loop area in the circuit. Can you put a steel screen between the traffo and the input socket? BE CAREFUL OF THE MAINS WIRING THOUGH. CONNECT YOUR STEEL SCREEN WITH A STOUT WIRE TO THE SAFETY EARTH ON THE TRAFFO FRAME, PREFERABLY BOLTED BOTH ENDS.
Edit: this is what I did in my headphone amp in my avatar to rid it of a slight hum. It uses a frame traffo too.
 
On the extreme off chance that the 2.2nF capacitor was a mega-receiver of hum, I replaced it with one scavenged from an old (rather fancy) Toshiba CD player. Made no difference whatsoever.

BugBear
 
I find it extremely odd that removing the long, unshielded leads running from the PCB to the jack plugs made literally zero measurable difference to the hum.
 
I note that I don't have a value for the tiny ceramic capacitor next to the 100kΩ resistor on the +ve input to OA2.

So I played with inspection mirrors and macro lenses for a while. It's 47pF.

This means that there is a 100kΩ resistor to virtual ground and a 47pF to PCB-gnd on that node.

A quick play with an impedance calculator gives me: 100Hz-33MΩ, 1KHz-3MΩ, 10kHz-33kΩ for that capacitor.

I don't understand what this resistor/capacitor are for.
 
The 100k performs two functions, it is a load on your source and provides a tiny bias current to that imput. The 47pF cap limits bandwidth as it shunts hf/noise to earth.
 
For diagnostic purposes, I have done the opposite of trying to make the earth hum problem go away.

I have attached (unshielded) leads 29cm long to the 2 inputs, and joined them together, and thence to PCB-gnd via a double ended croc lead of 24 cm.

bad_leads by plybench, on Flickr

Input leads are pink/gray. The little wire loop is bare wire soldered to the PCB-gnd track.

So what did effect did this appalling wiring have? The +ve input of OA2 was at 1-2mV, 50Hz, and the output at 27mV.

In other words, exactly the same as the shortest wire straight to GND.

I tried moving the wires around, including over the PCB, transformer - the voltages on OA2 stayed the same.

My confusion continues.
 
After all the above, the only thing I can think of is if the 100uf connected between the 0V plane and the virtual earth is either a bit shonky, or is slightly too high an impedance at 50Hz, you are getting a tiny bit of ripple on your virtual earth. As your inputs and power amp sections reference the 0V plane whilst the opamps reference virtual earth, any ripple between these two will be spread throughout the entire opamp section. Can you put say 220uF on the back of the circuit in parallel with the 100uF, i.e. between 0V and virtual earth? 100uf is about 32 ohms at 50Hz whereas 320uF is about 10 ohms, so if there is any slight ripple, it should be reduced with this in place (fingers x'ed.)
 
a.palfreyman said:
After all the above, the only thing I can think of is if the 100uf connected between the 0V plane and the virtual earth is either a bit shonky, or is slightly too high an impedance at 50Hz, you are getting a tiny bit of ripple on your virtual earth. As your inputs and power amp sections reference the 0V plane whilst the opamps reference virtual earth, any ripple between these two will be spread throughout the entire opamp section. Can you put say 220uF on the back of the circuit in parallel with the 100uF, i.e. between 0V and virtual earth? 100uf is about 32 ohms at 50Hz whereas 320uF is about 10 ohms, so if there is any slight ripple, it should be reduced with this in place (fingers x'ed.)
Click to expand...
1) My longer term plan was to replace ALL the electrolytics any way.

2) Since modern caps are smaller than old caps, AND 220µF only came in 5 packs (needed for the pre-amp 25V power rail) I was going to use 220µF for the virtual ground as well.

3) In general I do not have stocks of spare parts (since I am not really a DIY constructor), although I am merrily scavenging an old Toshiba CD player, one of these, $1000 in 1983:

http://www.thevintageknob.org/toshiba-XR-Z90.html

Do you recommend moving ahead with my re-cap?
 
I certainly won't hurt to replace the 1000u main PSU caps and the 220u + 100u that provides the supply and the virtual earth respectively of the opamps. By all means replace the 100u with one of your 220u caps.
 
You must log in or register to reply here.


advertisement


Back
Top