OP amp oscillating/modulating OP.

[Diabolical Artificer]

I'm working on a powered sub for a mate and have got as far as U3, see att. All is well at pin 5, but the OP at pin 1 is modulated by 100hz. I've powered this up on various PSU's, same result.

What should i look at, what could be the cause? The caps are mostly poly types, IE C3, or could it be the fets T3/13 causing an issue?

Advice welcome. Andy.

[Diabolical Artificer]

Update, T13 was SC between G&D and I now have HF oscillation on U3 pin 2 IE 666,666khz (1.5 divs TB set to 1uS). The thlot plickens. I wonder if U3 is damaged? Bit reluctant to whip it out as it's a double sided board, easy to cause damage.

A

[duncanlowe]

Originally Posted by Diabolical Artificer

Update, T13 was SC between G&D and I now have HF oscillation on U3 pin 2 IE 666,666khz (1.5 divs TB set to 1uS). The thlot plickens. I wonder if U3 is damaged? Bit reluctant to whip it out as it's a double sided board, easy to cause damage.

A

One has to wonder what caused T13 to die in the first place and if the oscillation is related to the cause, or is caused by related damage.

[Diabolical Artificer]

I can't find any other damage related to T13. The main power amp functions ok. I've tested a few components that connect to U3, all seems ok. I presume U3 is a LPF and wondered if the LF modulation is part of it's function. I've been testing the preamp with a 1k signal, not a good idea I think. At 300hz though the signal is still modulated.

I've seen video's of other amps where J113's are employed in similar circuits and have died. I got the impression that they were kind of like Lemmings.

I've attached the rest of the schematic relating to U3 and it's power amp bit too.

Andy.

[Radio Wrangler]

T13 is DC coupled to the LF output on the multiway connector. Someone only has to apply something with some stored charge or ESD and it can b*gger the FET.

THere doesn't have to be an internal reason.

David

[Argus25]

Since U1 has a part of its circuitry connect to U3, and there is a feedback loop around U1 that involves a 100uF filter electrolytic, C11, one of the things to check is if that electro has gone low value or high ESR, it could possibly result in oscillations that end up appearing in U3's output.

[Diabolical Artificer]

Ok, had a look at C11 and also C6, C11 checks ok for ESR, C6 showed higher ESR but within specs, removed it anyway. Also lifted one end of C3 and C2 and also R31. The only thing that stops oscillation is the lifting of one end of R24, the 4k7 FB resistor, this kills osc stone dead.

I've tried about everything I can think of apart from changing U3, will see if I can do this without damaging the PCB.

Andy.

[Diabolical Artificer]

Changed U3, oscillation gone, so a duff opamp. I didn't have a 4558, which is what was fitted, tried a 4565 and another similar type, ended up fitting a TL072 which works best. I put a socket in as the original opamp was a nightmare to get out, only damaged one part of a track luckily.

So aside from ordering a few J113's and maybe a few 4558's the jobs a good un. thanks for all your input.

Andy.

[Radio Wrangler]

It looks to be marked NE5532 in the schematic posted up-thread.

NE5532 is quite a high performance opamp... A dual NE5534... positively frisky.

It's unlikely that a 4558 jellybean 741-like amplifier would make a good substitute, it will have too much phase shift coming in at too low a frequency.

David

[Diabolical Artificer]

They're all 58's in there David, so must have been fitted as factory. the heading on the schematic says "prototype release" if that means anything.

Andy.

[Ted Kendall]

They may have had stability problems with 5534s - Self, who uses 34s and 32s a lot, finds that 10nF between the two supply rails (not between rails and ground), as close to the chip as possible, is helpful in preventing instability.

[Radio Wrangler]

Originally Posted by Ted Kendall

10nF between the two supply rails (not between rails and ground), as close to the chip as possible, is helpful in preventing instability.

That's a classic trick with faster opamps. Opamps have no direct ground connection and only know what ground is via input and feedback networks. supply-to-supply impedance is more important, so a single C from supply pin to supply pin saves stray L and gives a lower risk of the thing hooting.

5534/5542 are quite tame, given good housekeeping around them. Frisky. Some of the Comlinear current-mode feedback parts were positively mental, though.

If you're using much of the 5534's rather good output current capability, then rail to ground decouplers come into their own.

David

[Craig Sawyers]

The RC4558 datasheet recommends 0.1uf low ESR capacitors between each supply and ground, as close to the device as possible. It says it is electrically similar to the 741, and although it has better noise performance, it unity gain bandwidth is only 3MHz and slew rate is 1.7V/us, which is pretty sluggish.

But an 072 is a fet input op amp and the 4558 is bipolar, so there are electrical differences.

[Radio Wrangler]

It looks like some of those opamps pass signals to the main amplifiers and main speakers, so I'd prefer something a bit better than 4558s.

With standard basic pinouts being common for single and dual opamps, I can see people 'tube-rolling' the things sometime in the future. Blissfully unaware of massive differences in bias current, need for external compensation, etc.

David

[joebog1]

Originally Posted by Diabolical Artificer

Changed U3, oscillation gone, so a duff opamp. I didn't have a 4558, which is what was fitted, tried a 4565 and another similar type, ended up fitting a TL072 which works best. I put a socket in as the original opamp was a nightmare to get out, only damaged one part of a track luckily.

So aside from ordering a few J113's and maybe a few 4558's the jobs a good un. thanks for all your input.

Andy.

Easiest way to not ****** the PCB is to cut all the pins off the chip with sharp sidecutters and remove each one separately. I dont have a "solder sucker" either. By that I mean an electric pump with a hollow soldering iron tip.

Cheers Andy

[Craig Sawyers]

Indeed David. The faster the opamp the more finicky they are regarding layout and decoupling. So something that works with slower opamps might be a good oscillator with a modern one.

Although it is only SOIC with no DIL version (although adaptors are cheap enough), the generic pinout applies to this http://www.ti.com/lit/ds/symlink/opa1656.pdf , which with 53MHz GBP and 24V/us slew rate. In fact the only questionable spec for this animal is the 1/f corner at about 1kHz, typical for JFET input op-amps.

However that sort of raw performance needs very clinical layout and decoupling to keep it in order.

Craig

[Radio Wrangler]

I've used the AD811 rather a lot in the past, that's a 2,500V/us part... as well as AD8351 more recently which hits 13,000 V/us and gives 2.2GHz bandwidth with the gain set to +12dB. Let's say layout around the inputs and the gain setting resistor (interleaved pins!) needs to be really tight.

Both these parts are actually very well behaved, provided sensible layout and component choice are done.

It's rather difficult to get people to understand that the sort of layout a part needs is not something appropriate o the frequency which it is being used at, but rather something appropriate to the frequencies it could sustain oscillation at.

David

[Argus25]

There is not much point in having the op amp much more peppy than what you need.

While the 741 for example is very slow and a little poor for HiFi at least it is totally deaf at RF frequencies and that , sometimes, is helpful.

For Hi Fi audio work there is little point in having the op amp perform above 200kHz if in closed loop conditions you only need 20kHz.

Super high slew rate OP amps for RF work do require a much more careful pcb layout and bypass cap position & quality. Though a part like an AD8056 is pretty tolerant to bad layout.

A very good type of OP amp for audio work is the LinCmos type, for example a TLC272. These are super well behaved , have a great rail to rail output swing with the two fets. Are capable, even at ultrasound frequencies around 40kHz. But they roll off around the bottom of the RF spectrum, perfect for audio really as they won't set any limit in closed loop conditions due to the IC itself and the limits are set by the external components as it should be. They are nice and stable and don't misbehave with any layout I have used. The internal design is also very elegant.

http://www.ti.com/lit/ds/symlink/tlc272.pdf

[Diabolical Artificer]

"Easiest way to not ****** the PCB is to cut all the pins off the chip with sharp sidecutters" Thanks for the tip mate, I remember hearing of that before but forgot about it. As well as this active woofer I have it's HF sister to look at with an iffy IP socket. Tried resoldering the pins on a combined XLR/1/4" jack socket, better but no signal on pin 3/tip.

So mad socket out, yep, double sided board with, oh happy days, six pins and big b*ggers too. Managed to get it out with a crowbar, but hoorah, damage to one side of PCB. BTW, cause of no signal was a 3mm long 100r SMD resistor slanted sideways on pads. Whilst trying to reposition it legged it and hid. Managed to get another off a scrap board (it is worth saving old router boards) only that one disappeared too. Third time lucky managed to solder back on, magnifying glass in left hand, solder in teeth, iron in right. Struth, I'm not even getting paid for this. Anyroad fixed broken tracks with sticky copper tape and prayed it worked, it did thank all the electron gods.

Anyroad, back to opamps. I ordered some 5532's as well as looking what I had on scrap boards, mostly 4580's, a few 4560's and a few other odds and sods. I've yet to fit a 32, but whilst waiting for bits to arrive I had an issue with a LPF I made using a TL072. I've found these sub really well for most opamps in PSU's and other applications, but on my LPF it was ringing. I've always treated opamps as triangles with gain, but reading the above replies I can see I'll have to choose the right opamp for the job, and watch layout and decoupling.

Thanks, Andy.

[Argus25]

I was interested when I built my TRF radio recently to do the audio sweep with the vintage 741 OP amp.

By 19kHz it slew rate limits, so the amplitude doesn't drop much but it does a beautiful job of turning the 19KHz sine wave into a triangle wave.

In fact one really good way to turn a square wave into a triangle or trapezoid, is to simply pass it through a low slew rate OP amp. It works just as well as a current source charging a capacitor and generates a beautifully linear ramp.