Intersound IVP layout

[jimladladlooklike]

I switched the DC Jack with a plastic one and it has sorted the weird beeping issue. I'm not sure but I think the previous DC jack was causing a short between the +18v and ground.

However, I'm now getting DC voltages on all 3 1/4' jacks.

[jimladladlooklike]

Hi again folks.

So turns out my power supply section was not what was required (I'd basically made a fancy voltage divider), but now I've made a bipolar supply which shares ground with the +18vdc in.

I think I'm finally getting somewhere, in that it works quite well. It's loud enough and has a nice amount of headroom. However, it's still noisy. After measuring the negative supply voltage I noticed that it fluctuated between around -14v and -16v. Could this be a filter cap issue?

I've uploaded the schematic I'm using below.

Cheers guys

[jimladladlooklike]

I should have mentioned that the reading I got was when the supply wasn't hooked up to load. I changed the inductor and now the negative voltage is around -15vdc and stable. However, the voltage drops to about -8vdc when loaded. I'm guessing the PS doesn't meet current requirements.

[MikeA]

Try replacing the 7809 with a second 7815, and replace the 637 inverter with a Microchip TC962 CPA, it handles +15v nicely and outputs -15v (or darn near) @ up to 80 mA.  The pinout is different than the 637 so watch that.

[jimladladlooklike]

Thanks Mike, will check that out.

[jimladladlooklike]

So my ammeter reads around 250 mA when hooked up in series with the circuit, so I may need another solution. I have some ideas of what I need to do so will report back when I can.

Cheers

[Rob Strand]

You need to measure the *output* current on the +V and -V rails.
[Probably clearer if I said you want to measure the current on each rail *to* the IVP circuit.   For the negative rail that's the output current of the converter.]

I estimate about typical 50mA on *each rail*.   Worst case is less than 80mA to 100mA.
The 43 ohm vs 430 ohm tube sound error on the schematic will pull more current.

The original power supply had a current limit of less than 150mA on each rail.

The DC converter is only say 75% efficient so the -V rail will pull about 67mA typ. (150mA max) from the +ve DC in rail.

Add that converter current to the +V current gives a total input current of 120mA typ  (250mA max).

If you converter isn't working very efficiently due to an incorrectly chosen inductor then that would show up
as an increase in the DC input current.

As far as your noise issue goes,  if you converted +18V to -18V (or above) then used a -15V regulator on the output of the converter it is likely that will reduce noise.   Alternatively use a DC converter without and regulator and place an RC filter or active filter on the output of the -15V to remove noise.

[jimladladlooklike]

Hi Rob, thanks for explaining that to me. Finally got round to taking said measurements.

The current on the +ve rail was 87.3 mA
The current on the -ve rail was 86.9 mA

They did start off much higher however (+ve around 120mA and -ve around 103mA) then reduced rapidly then gradually settling at the above readings)

So that means a total input current of about 180mA, right? Does this mean the 250mA I previously measured at the DC input shows that I've used the wrong inductor?

If it helps then the supply voltages are as follows:
+ve 14.95 VDC
-ve 13.25 VDC

The from the wall wart itself is +17.83 VDC

Cheers

[Rob Strand]

The current on the +ve rail was 87.3 mA
The current on the -ve rail was 86.9 mA

Based on the estimates I made those look a little high.   Not sure why.  I could have screwed up.   Nonetheless it might worth checking some stuff.    In a few places in the circuit is an opamp feeding two transistors and those transistors have 10 ohm resistors in the emitters.   Measure the voltage across each of the 10 ohm resistors.  You can then calculate the current through the resistors as I = Vmeasured / 10.

If we take your -ve rail current at 86.9mA and -15V output that's 1.30W.    According to the datasheet, and assuming all is well, the converter efficiency is 85%.  The converter input power should be 1.3/0.85 = 1.53W.   I realized you are feeding the converter from a 9V rail so that means the input current to the converter will be 1.53W / 9V = 170mA.   The input current to the 9V linear regulator will be roughly the same current, so 170mA  from the +ve rail going into the 9V regulator.

The total current expected from the positive rail is 170mA + 87.3mA = 257mA.    Which is in very good agreement with what you measured.

What that concludes is the converter is working correctly.

Having the 9V regulator at the input to the converter is throwing away half of input power because the 9V regulator drops 9V across it and produces 9V out (the 9V drop loses the power as heat).  The converter chip can handle 16.5V in but not with -15V out.  With -15V out the input voltage needs to be 24V - 15V = 9V.  So the chip is already max'd out.   In theory you can save power but not with that chip.

At this point it might be worth checking the current into the various sections of the circuit to see if there isn't a problem somewhere.

[jimladladlooklike]

Apologies, I should have mentioned that I removed the 9V regulator and supplied the converter with the output of the 15v regulator, the same rail as is supplying the rest of the circuit with v+. That's how I got the measurements shown. I did read the table on the datasheet, hence why I originally decided to supply the converter with +9v from the regulator, but I thought it might be worth a try supplying it with +15v. I did some calculations and decided on a 220uH inductor. (I think I did this right, but it took me a while to figure out, so who knows?)

Anyway, if you're saying the chip is maxed out when supplied with +9v, does that mean I could be damaging the chip supplying it with +15v, or that it will just make no difference to the outcome?

Regardless, I will check over the schem/layout/build itself and measure currents at the points you mentioned and report back.

Cheers again

[Rob Strand]

Anyway, if you're saying the chip is maxed out when supplied with +9v, does that mean I could be damaging the chip supplying it with +15v, or that it will just make no difference to the outcome?

The way I interpret the note in the datasheet it could  be damaging the chip.   My guess is the voltage limit is due to the MOSFET ratings inside the chip.

It's probably just a fluke +15V current measurements match  the calculations for +9V.   If the chips isn't doing something weird and just holding onto it's life then the current does seem high for +15V in and then perhaps something isn't quite right with the inductor.   (For example if the inductance value is on the small side it could be saturating and that might push-up the input current.)
 

[jimladladlooklike]

Hi again!

So I took a break from this project, partly as I needed to build some stuff for friends and partly because I think I destroyed the chip... It got insanely hot then the thing just stopped working.

Anyway, I found a power supply that I thought might work here:

https://www.ebay.co.uk/itm/303536235236?ssPageName=STRK%3AMEBIDX%3AIT&_trksid=p2060353.m1438.l2649

I hooked it up and got the circuit running at +/-15V, no problem. Once again, it sounds good, all controls work well, but there's a different noise this time. A buzzing that starts on power up, decreases in pitch for a few seconds before settling and continuing at an annoyingly loud volume until powered off. It's not like a hum I've ever herd before, it's very buzzy.

[jimladladlooklike]

I again took current readings again for both +ve and -ve rails, same results as last time give or take a mA or two, and I didn't let them "settle for as long as I did last time.

I also took readings across those two 10ohm resistors on the emitters of the trannies you suggested. both measured 0.007VDC, so 0.007/10 = 0.0007 so the current through the resistors is around 0.7mA. Does this seem good?

Cheers

[Rob Strand]

Switching supplies can pulse under light loads which can cause whining.  That power supply is 20W and the chip perhaps can go even higher.

Try putting some dummy loads on the power supply outputs.   Start with say 220 ohm.    Given the power supply rating you might need to try 100ohm or 47ohm.    For the lower values you will need to increase the resistor wattage or use parallel resistors.

Beyond that would be filtering.

[Rob Strand]

I later remembered the details of this thread.   The circuit was pulling a reasonable amount of current already.   Nonetheless the dummy load is still worth trying.

I also took readings across those two 10ohm resistors on the emitters of the trannies you suggested. both measured 0.007VDC, so 0.007/10 = 0.0007 so the current through the resistors is around 0.7mA. Does this seem good?

That seems OK.   

[jimladladlooklike]

Okay, so start with 220 ohm 1/4w carbon film resistors? Then maybe 100 ohm 1/2w?

Cheers again

[Rob Strand]

You can work it out from P = V^2 / R.

For 15V,

R          P [W]       Current [mA]
470      0.5            32mA
220      1.0            68mA
100      2.3            150mA
47        4.8            320mA

If you only have low power resistors you can use a number of higher values in parallel.

From your previous measurements you circuit was pulling about 87mA for each rail.

Typically we might expect a switchmode to work OK at 10% of the rating so 2W which is 68mA per rail.  However when it comes to audio that rule of thumb can quickly go out the window.   If you put 2x470ohms 1/2W resistors in parallel on each rail it will increase the current draw by 2x32mA = 64mA.

If you hear a change in pitch of the whine then you know you are on the right track.   If the 64mA isn't enough to remove the whine you will have to add more resistors.   The question is how much load is required to get rid of the whine, and can you get rid of it at all?   You can only do this by experiment.

I noticed the advert says:
"I've measured ripple at approximately 200kHz at 200mV peak to peak when driving 12V at 300mA."

So you might need to add some RC filtering on the supply anyway.

[jimladladlooklike]

So I put 2 x 470ohm 1/2W resistors in parallel from V+ out on the power supply to V+ in on the circuit, the same from V- out on the PS to V+ in on the circuit.

The whine got lower in pitch. I used the trim pot on the PS to reduce the output voltage and after s couple of turns the whine disappeared. If I remember rightly the output voltage on V+ and V- rails sat at around 14.3V and -14.3V. (Ish). The problem now is that the volume has decreased by a significant amount, not reaching unity gain with a signal going direct from guitar to amp.

ALSO, me being the idiot I am, I wanted to see what would happen with just 1 x 470ohm 1/2W resistor on each rail, so I hooked those up and I either shorted something with the crocodile clips or the circuit just didn't like it? There was a high pitched squeal then no sound except a steady regular popping sound.

The voltages were jumping all over the place after this occurred. So I disconnected the PS from the circuit and they're steady, so I guess I've messed up something in the circuit itself. Think I'm going to try and replace every active component when I get chance. Hopefully that will sort said problem out.

Following that, I will try and find a combination of resistors that rid the whine...

Thanks again for your help

[Rob Strand]

So I put 2 x 470ohm 1/2W resistors in parallel from V+ out on the power supply to V+ in on the circuit, the same from V- out on the PS to V+ in on the circuit.

The whine got lower in pitch. I used the trim pot on the PS to reduce the output voltage and after s couple of turns the whine disappeared. If I remember rightly the output voltage on V+ and V- rails sat at around 14.3V and -14.3V. (Ish). The problem now is that the volume has decreased by a significant amount, not reaching unity gain with a signal going direct from guitar to amp.

You might be close to the edge of it working.  Probably a good idea to add some more load resistors when the thing is on you desired voltage.

ALSO, me being the idiot I am, I wanted to see what would happen with just 1 x 470ohm 1/2W resistor on each rail, so I hooked those up and I either shorted something with the crocodile clips or the circuit just didn't like it? There was a high pitched squeal then no sound except a steady regular popping sound.

These things happen.    High pitch-squeal and on/off (which would cause popping) is a sign of a short.

The voltages were jumping all over the place after this occurred. So I disconnected the PS from the circuit and they're steady, so I guess I've messed up something in the circuit itself. Think I'm going to try and replace every active component when I get chance. Hopefully that will sort said problem out.

Yes sure sounds like short or fried part on your board.   You can narrow it down by remove the power *both* +ve and -ve to sections of the of the board.

[jimladladlooklike]

Okay so I've got it working great, no whine either. Some relatively quiet hiss, as expected, but it's finally useable.

The problem was a building error, which I will get to.

I narrowed down the problem by audio probing and taking g voltages at various points. Turns out there were 2 x ICs that had been destroyed during the previous incident. Once those were replaced I continued with the audio probe and discovered that the whine was coming from the 1st output (pin 1) of the last 4558 in the signal chain. Also noticed that said IC was running pretty hot. So, looked around for a bit and discovered one of the 13K resistors, specifically the one going from +15v to the base of the 2n4401, was in fact 13ohms. Once I replaced that the preamp worked perfectly.

Finally! Turns out I just needed to check my work more thoroughly. However, I wanna thank you for helping me out. I've learned some skills I've never had to use before, which will surely help me in the future.