Debugging the FM driver

[hans h]

Hi all,

I've had a tremendous amount of help here already, but ran into trouble with one of my own builds again 

I call this thing the FM driver, which is something like a heavily modified stacked honey bee and dyna red tuned for a dual rectifier like frequency response and with addition of a tone stack, boost and presence control.











I get signal up to the +input of q3 and very little afterwards. The voltages are also off for q3 and q4 (tl072, for voltages see picture). Any idea what might be causing this? The vref seems good. Q3 is a very simple buffer.

Edit: since this schematic was my own doing there may be three general causes of the problem:
1. The opamp is broken or latched-up.
2. The schematic is flawed.
2. Something went wrong with soldering.

[antonis]

I think you have to seperate Vref from GND..

[hans h]

You mean: please do not use the symbol for ground at the vref points? Sorry, schematic was drawn in ltspice and for frequency purposes I used ground in ltspice.

In real life it is +9.9v - > series diode - > 100r - >100 uf filter cap-> 10 k 10k divider with 33uf filter cap.

Edit: vref is then tapped to the +inputs of the various opamps using resistors of various values. Vref does not seem to be the problem as the voltage at the + input of q3 is correct. (4.35V)

[hans h]

As another aside: the guitar signal at the + input sounded delicious, so I was really bummed that something is up with q3/q4. If I cannot get it to work, I suppose I can always snip off the last part of the circuit 

[PRR]

....schematic was drawn in ltspice and for frequency purposes I used ground in ltspice....

I realize this is not your problem, but:

LTspice can have "Named nets" other than the universal 0 GND. It is that box with an "A" next to the GND symbol menu.

Also we can screen-grab a schematic and get a clear image.

[antonis]

Vref does not seem to be the problem as the voltage at the + input of q3 is correct. (4.35V)

Never said that..

I'm just trying to read your schematic so, are there any Vrefs other those marked so arrows..??

P.S.
I'd check U4 pin 1 for short to GND..

[hans h]

Thanks PRR and Antonis. From now on I'll use the named nets (with ground to boot if necessary). I'll take clearer pics and check q4 tomorrow.

[hans h]

Below I attached a clearer schematic. However, I wasn't able to attach a name (text) using the "label net" button.

Antonis, you were correct: something fishy is up with the Q4 + input or thereabouts. There is only 0.7 k ohm separation from ground. I'll be looking into the soldering around that point, thanks!

I was so focussed on Q3 because that's the first part where I lost signal, that I completely overlooked the low voltage on the Q4 + input 

[hans h]

Okay, I am officially clueless now. Q4 +input is only connected with two components: 330nf coming from the treble pot and a 330 k resistor going to vref. I broke the connection between + and both components.

Then I re-measured the resistance between + and ground: 0.97k ohm, exactly the same as before! Does this mean the opamp is broken?

[anotherjim]

Depends on the polarity of the meter probes between +input and ground which is the -V supply pin of the chip. There can be diodes in the chip that will not conduct in normal use where the +input is always more positive than the -V supply but will conduct with the polarity reversed. The meter puts out a positive voltage on the red probe to do resistance or diode/continuity tests. Note that good old analogue moving needle meters have reverse polarity (-v on the red probe) for resistance ranges.

[anotherjim]

Oh, and you can't read resistance if the board is powered and even with the red probe on +input it may still find a path around to the +v supply.

[hans h]

So if I understand correctly: for a digital meter, red to + input and black to ground should give a high resistance (what value is to be expected?

[anotherjim]

Well, R15 goes to Vref which is a resistor divider and I'd expect it to be dominated by the value of the resistor to 0v so that is in series with R15. Any capacitors on Vref to 0v will take a little time to charge up to the meters test voltage so expect the reading to start low and rise up to the true value.
Ideally, the IC is in a socket so with that out you can ignore the possibility of a faulty IC until you prove the +input components are ok.
Other things to look for...
I don't know what IC packages you have - single, dual or quad - but they all have at least one +input next to the -Vsupply pin which is 0v in this case so a short between those pins is possible.
Easier to happen with a single opamp package, the feedback link from output to -input is routed to the +input in error.

[hans h]

The funny thing is that the reading remained the same after breaking the connection between opamp + input and the 330k that goes to the 10k 10k voltage divider.

The suspect opamp is the 2nd half of a tl072. Unfortunately my stock of opamp sockets was depleted so I soldered the opamp straight to the board.

[hans h]

Right now I think the only thing that can be causing trouble is the opamp itself: out-of-circuit the 2 components attached to q4+ are in spec. At the moment, nothing is connected to q4+. Resistance to ground is 0.97kohm both with black and red lead of the multimeter. Are there any other possible culprits? Otherwise I'll yank the thing out.

[anotherjim]

I would change the opamp at this point.

[hans h]

Thanks! As soon as I get time I'll do that and post back the results.

[hans h]

So I installed a socket for the opamp and tried another tl072. The result was exactly the same

I am not as patient as I should be for this kind of stuff, so decided to ditch all circuitry from the coupling cap towards q3 onwards. I added a simple jfet buffer to the end and hooked that up to the volume pot. The circuit is now working great but obviously I haven't got the bass and treble controls anymore.