Is my scope misbehaving?

[Halvmand]

Last night i was playing with an amplifier I made last week. Its made from a lm 386 chip, with a gain of 200.
This was mostly for trying out some of the functions of the scope i haven't used before.

So to see how much the the outgoing signal differs from the input, I used the math functions to subtract the output from the input.
All fine and dandy, though the difference between the signals persisted.

So to be sure that it was the amp and not the scope or probes that made the resulting waveform, I decided to probe the input with both scope channels.
I was expecting a straight line on the resulting waveform. or at least much less deviation between the two.

As you can see on the first picture, the resulting waveform (bottom one) is not straight at all, and Vpp = 500 mV.

Scope was set to 4V/div on both channels and no offset.
X10 probes was used
Frequency was 11,8 kHz

So the question:
Are there something wrong with the scope, the setup, the probes............... or me.
I suspect something with the scope probe compensation?

[Halvmand]

Probes is within 5 mm of each other.

[T3sl4co1l]

Can't make out what the probes are connecting to.  Why is it significant they are within 5mm?  Schematic?

If you are measuring Vo - Vin, and Vo = 200 * Vin, are you not simply measuring 199*Vin, which looks enough like 200 not to notice the difference???

Of course, that wouldn't explain your apparently having 0.5V input.  But that begs the question, why are you driving it with 0.5V when the amplifier is utterly incapable of producing 200 * 0.5V = 100V output?  Surely it should be clipping to a voltage somewhat below supply?

I don't know enough about that HP scope to tell what traces are what, at what scales, etc.  They don't look to be labeled.

Also, are you using 10x probes (or not), and is the scope reading them correctly (auto-detecting, or manually configured)?

Tim

[Halvmand]

Probes are connected to the output from the signal gen right at the amps input. 5mm from each other on the same wire. thought it would matter slightly because of the very small voltage drop that might occur if they where different places.

Vpp at the probes is about 3 V. no signal to the amp since the pot was turned down.
Sure the amp would clip at .5 v input. Thats why I connected the scope and turned the pot until i got an output without clipping.
This worked fine.

I might have confused you by having the amp in the picture at all. I just snapped the pictures of the setup as it was.

From the top:
Channel 1. 4V / div
Channel 4. 4V / div
channel 1 vs channel 4
channel 1 minus channel 4. 500 mV / div

I am using two x10 probes. and the scope is configured manually to 10:1

[T3sl4co1l]

Oh, so the difference isn't on scale at all, it's magnified by a factor of 8?

So your probes are reading the identical signal (not input and output) and one is reading about (1 - 1/8) = 88% of the other?

Probe compensation?  Might be a part of it.  If you don't have a voltage reference, you wouldn't really be able to calibrate those probes against anything.  They do need to measure the same at AC as DC (so the square waves are as square as 100Hz as at 100kHz), but I wouldn't assume that toy of a signal generator would be clean enough to use for that.

Tim

[Halvmand]

I rearranged the scope setup and it is now more obvious where the error is.

I've attachet a picture. That's a badly compensated probe right?

I don't have a voltage refference, but they come cheap right?
You are probably right, that the signal generator can't be used for probe compensation. Rise time is slow and it only goes to 60 kHz.

[Dave Turner]

What model scope are you using? Does it have an output signal for probe compensation?

[Halvmand]

It is a HP 54501A. No output for probe compensation on this one.

[tautech]

I rearranged the scope setup and it is now more obvious where the error is.

I've attachet a picture. That's a badly compensated probe right?

I don't have a voltage refference, but they come cheap right?
You are probably right, that the signal generator can't be used for probe compensation. Rise time is slow and it only goes to 60 kHz.

Probe compensation os usually done @1 KHz and ~3-5V squarewave, set your SG for that.

It is a HP 54501A. No output for probe compensation on this one.

Yep just looked at online images, it doesn't appear to. 

Swap probes as a check for error once compensation is adjusted.

[bson]

Looks like there's a negative offset on the top waveform.  This offset*gain is probably what you're seeing in the difference.

[rolycat]

It is a HP 54501A. No output for probe compensation on this one.

Yep just looked at online images, it doesn't appear to. 

The HP 54501A does have a probe compensation output, but it's a weird one - a 1.5 MHz signal output to a BNC socket on the rear panel. A probe to BNC adapter was used to connect the probe.
 

[tautech]

It is a HP 54501A. No output for probe compensation on this one.

Yep just looked at online images, it doesn't appear to. 

The HP 54501A does have a probe compensation output, but it's a weird one - a 1.5 MHz signal output to a BNC socket on the rear panel. A probe to BNC adapter was used to connect the probe.

Yes that is a weird one, wouldn't you agree?
I stand corrected, Thanks.
Do you know the amplitude?
It would make sense to route it to the front with a BNC cable, maybe with open ends(loops) for easy use/accesibility. 

[Halvmand]

It is a HP 54501A. No output for probe compensation on this one.

Yep just looked at online images, it doesn't appear to. 

The HP 54501A does have a probe compensation output, but it's a weird one - a 1.5 MHz signal output to a BNC socket on the rear panel. A probe to BNC adapter was used to connect the probe.

You bastard. Now i feel like an ass.
Tomorrow I will look in the manual, and hopefully there will be a section about probe compensation 

And, thanks for all the answers guys. I'll keep you posted.

[Halvmand]

I've tried to compensate the probes today and the whole situation now looks much better.

The first picture is with the in build square wave generator. Again, channel 4 is subtracted from channel 1.
Both channels are at 600 mV / div and the output function is at 100 mV / div.

Second picture is with the Chinese one.
Both channels at 4 V / div and the output function at 500 mV / div.

bson: You might be right regarding the offset. It still persists, but is smaller than before.
Would it be possible to hunt it down and correct it, you think?
I've tried to analyze what is causing it.
By analyzing I mean doing the math on paper and compare it to what's on the screen. Se the third picture.  1 - x looks a lot like what i see on the scope.
My brain says it is a mix of both probe attenuation and some offset?

[Halvmand]

Third picture.