VOLTCRAFT DOV series

[SL75]

I experimented a bit today.
The 5.3V supply has some peaks, but the 5V is totally clean.
(Attachment Link) (Attachment Link)
The 2.5V reference has the notorious 7.6kHz, which vanishes if disconnected (remove 1Ohm).
-> So the 7.6kHz comes from the load connected (oscillation) and not the ref voltage or buffer itselt.
(Attachment Link) (Attachment Link)
I then realized that with the SL75 mod, the signal is a bit worse than when removing the opamp alltogether, so I´ll stick with the zero ohms.
(Attachment Link) (Attachment Link)
I also did the re-route of the reference to the OPA145 for each channel, which I think improved the signal a bit at 200mV.
Did not work. Found out that it shifts vertical offset....

I´ll leave it at that for the moment. Maby someone else has more improvements.

Edit: Also "replaced" the battery which had only 2.7V (fresh delivered unit) and did loose date and time.
(Attachment Link)

I also tried the sense line of the REF3425. Re-route it to after the buffer. That results in some really funny oscillations 

The thing with the lithium battery could be dangerous, because it seems like Owon soldered a rechargeable lithium cell from Japan here (and not a non-rechargeable standard cell from China).
https://www.fdk.com/product_e/lithium_e/coin_mnli/#ml
These cells are not safely interchangeable with each other. Maybe your cell was just not sufficiently (pre-)charged and that's why the voltage was so low?

[goaty]

Battery:
Good point. I have not measured if voltage increases during power on..
Edit: It's charged only when scope is powered on, not just plugged.

Reference Voltage:
From the schematic it goes to four/two frontends and there to the OPA145 and through a filter to the SGMOP17.

[Aldo22]

Mostly experience. Use a blob of solder over the resistor and it comes off at once.
Grab it with tweezers if it sticks to the iron.
Also use solder wick and flux.
I used leaded solder first on the opamp then hotair.
Most important is to have flux and isoprop alcohol for cleaning.
Magnification is also important if you´re over 50.
Either loupe with 3x (not 3 diopters! 3x is 8diopters) or better stereo microscope.

Dont need fancy  1000€+ Weller for that

Thanks for the practical tips! 
The tip about lead-based solder might help.

Of course, I’ve already watched videos on the topic where they use tons of flux.
But I don’t quite understand why that’s so important when desoldering with hot air.
Yesterday, for example, I removed a Windbond memory chip completely dry using hot air.
The component looks like new, and the board stays clean.
But here’s what I’ve learned: Never pull on the component with tweezers too early.
One pad came loose, but that’s hardly related to the heating method.
Other than that, I don’t see any collateral damage to the PCB, and I didn’t even shield the other areas from the heat.

Does how well this works depend heavily on the PCB?
My practice PCB (SAT-Rec), for example, loses its pads very easily.
With the red PCB, I think you could solder components on and off a hundred times without anything happening.

How delicate is the DOV PCB?

Thank you!

P.S. Sorry for the dumb questions, but I only have one chance to test this on the actual board. Better to make a fool of myself than end up with a broken oscilloscope. 

[Aldo22]

The thing with the lithium battery could be dangerous, because it seems like Owon soldered a rechargeable lithium cell from Japan here (and not a non-rechargeable standard cell from China).

Very good observation.
That might also explain why there isn't a battery holder. 

[goaty]

But I don’t quite understand why that’s so important when desoldering with hot air.

Does how well this works depend heavily on the PCB?

How delicate is the DOV PCB?[/b]

PCB type and quality varies extremely. You come across motherboard or mobile phone repair is for experts only.
You have 16 layer sucking away heat like nothing. Then you cannot solder or use hot air but need pre-heater.

OWON PCB is very friendly. I think I only slightly damaged pad of Battery which I probably falsely removed, because it is rechargable (akku).
I removed and re-seated the opamp and the resistors now 5 times, and its not damaged.

For remove using flux with hotair I think its reducing surface tension maby, but it´s not strictly necessary. What eases is reduce melt point by adding leaded solder.

FFT of scope:
I just experimented with FFT and I like it quite a bit. The peak measurement is pretty exact in values.
Trying with 1MHz sine and HP step attenuator shows exact dBV values (attenuation) down to less than 1dB exact.
Also I did not know it need not trigger when using FFT. At -80dBV it still shows peak, but cannot trigger (100µVRMS)

[Aldo22]

PCB type and quality varies extremely. You come across motherboard or mobile phone repair is for experts only.
You have 16 layer sucking away heat like nothing. Then you cannot solder or use hot air but need pre-heater.

OWON PCB is very friendly. I think I only slightly damaged pad of Battery which I probably falsely removed, because it is rechargable (akku).
I removed and re-seated the opamp and the resistors now 5 times, and its not damaged.

For remove using flux with hotair I think its reducing surface tension maby, but it´s not strictly necessary. What eases is reduce melt point by adding leaded solder.

Thanks! That's exactly the kind of information I was looking for. 
After all, it doesn't help to practice on a training object only to be confronted with a completely different reality when the situation gets serious.

[goaty]

Only thing that maby can happen is to destroy resistor or maby if careless opamp, if temperature too high. I have not more than 360.
Small resistor can be damaged the top resistive surface if scratch too much. Then you need replacement resistor kit from aliexpress (0603).

Also Aliexpress has good solder test kit for practice. I have order like 5 different even with BGA for very cheap. (Now more expensive thank you EU)

[Aldo22]

I have not more than 360.

Hot air gun or soldering iron?
I have the hot air gun set to 400°C and the soldering iron to 350°C.

Small resistor can be damaged the top resistive surface if scratch too much. Then you need replacement resistor kit from aliexpress (0603).

Yes, I need the 0603 components anyway, even if it's just for the 0R atm.
They've already arrived at Swiss customs. 

[goaty]

I have not more than 360.

Hot air gun or soldering iron?
I have the hot air gun set to 400°C and the soldering iron to 350°C.

360degree is soldering iron (temp controlled station)
Hotair need higher usually. I have around 450 with variing airflow.

For really small components, this kit is fun: https://www.ehajo.de/products/ise-100-incredible-solder-excercise
(Not smaller than 0201 though)

[Aldo22]

Beyond that, I have a more serious problem saving waveforms:
the images are saved without problems, but when I save the data (.csv, .mat, .zip), the files contain a correct header, but with truncated or incorrect data, or voltages that are constant at zero or a random value.
Has anyone encountered these errors when saving?

It actually works for me.
Attached: Mem depth 1M / 2Ch.
What firmware do you have?
What's the memory depth?
Maybe there's a problem with the flash drive?

[Martin72]

Dont need fancy  1000€+ Weller for that

Not for that, that's true.

[Olex]

Dont need fancy  1000€+ Weller for that

Not for that, that's true.

I've been doing great with a cheap USB-powered soldering iron for about 4 years now.
https://pine64.com/product/pinecil-smart-mini-portable-soldering-iron
As far as I know, there are already more modern and attractive options available on the market now.

[Martin72]

I made the modification to the buffer and took two measurements at the output—one before the modification and one after.
I almost ran into trouble during the modification:
Of all things, the 10k-ohm strip was missing from my 0603 resistor set from Amazon, even though it was supposed to be included...
But there were 20k resistors there, so I connected two of them in parallel and was able to continue.
On top of that, the resistors were hard to solder at first; I suspect there was some kind of patina on the terminals that had to be burned off first.

The result speaks for itself: there’s virtually no trace of the 39 kHz interference left.
The 7.62 kHz signal has increased slightly (+6 dB), but that’s a different issue.

[goaty]

I made the modification to the buffer and took two measurements at the output—one before the modification and one after.

The result speaks for itself: there’s virtually no trace of the 39 kHz interference left.
The 7.62 kHz signal has increased slightly (+6 dB), but that’s a different issue.

That is the SL75 mod with 10k and 5k1, right ?
So same effect I had. Ultimately removed the buffer altogether which results in lower peak.

Interesting measurement results. So it seems that some part of the circuit is drawing energy from the reference voltage source at a rate of f≈ 7.6 kHz. 
What do we know about this frequency? 
• It seems to be controlled by the firmware, because it doesn't appear during booting. 
• It seems to be crystal-stable, or derived (divided) from a crystal-stable source. 

Maby it´s caused by some internal switching/sampling frequency that loads/unloads the reference voltage in some way.
I think I have to measure and extend the schematic of the frontend a bit.

[Martin72]

That is the SL75 mod with 10k and 5k1, right ?

Yes, I wrote that in the picture, too.
I'll continue researching based on this mod, because, as I mentioned before, I think the buffer itself is a good idea for “decoupling” the 2.5V reference from the loads.
Here are two more DOV screenshots after the mod—it looks great, and you could leave it at that if you didn't know there was something else going on....

[goaty]

I wonder if adding some ultra-low ESR capacitance like 220µF POSCAP would migitate the ripple or cause more problems.

[SL75]

Regarding the remaining residual ripple at f ≈ 7.6 kHz, you could try further reducing the output impedance. For this, you could, for example, temporarily connect a 100 µF capacitor in parallel with C2329. The modified buffer circuit can handle that. The REF3425, maybe not... 
Maybe you have one of those ultra low ESR polymer aluminum solid electrolytic capacitors on hand. That would definitely be optimal here.

[Martin72]

What I find interesting is how constant the frequency is—all the others varied slightly, but the 7.62 kHz is “rock-solid.”
As we know, it only appears after booting up and stays there even if you close the Scope app.
@goaty:
That’s definitely related to something in the frontend, no question about it.
Shortly after the boot process, something is switched there.
I’d already started tracking the circuitry there again.
Oh, and even though no further proof is needed, I measured the 2.5 V again a little while ago.
Once before the buffer (i.e., the direct reference—green in the FFT)—and then after it.

[Martin72]

Regarding the remaining residual ripple at f ≈ 7.6 kHz, you could try further reducing the output impedance. For this, you could, for example, temporarily connect a 100 µF capacitor in parallel with C2329.

I had just connected some 100 µF capacitor from my scrap box to my measurement output (so not directly).
This causes the 7.62 kHz signal to be reduced by -13 dB (from -64 to -77 dBV), while multiples of that frequency remain almost unaffected.

[SL75]

I think that's a definite improvement. Since a standard aluminum electrolytic capacitor isn't low-impedance enough at higher frequencies, it can't really be effective there either. 
The main question still remains: What the hell is drawing such large pulsed currents (with f = 7.62 kHz fundamental frequency) from the poor reference voltage source?

[goaty]

Just added a 220u directly to it and I cannot measure the ripple anymore.
It´s gone !
(Panasonic POSCap has low esr)

The 'battery' really is charged... I put it back in... Sorry.
->It is only charged when scope is switched ON !

[Martin72]

@SL75:
That's exactly it, and that's why I'm going to continue investigating the circuit, regardless of whether a large capacitor reduces the symptom.
The 2.5V doesn’t just go to the op-amps (one of which acts as an inverter to convert it to -5V) in the circuit I’ve already drawn, but also to at least one other path, where I first detected it.
I’ll take a closer look at that path, because a logic IC is powered by it there.

Edit:

Just added a 220u directly to it

Right at the reference output—that's quite a feat.

[Arek]

Hello, I'm having a problem with my VColtcraft. From the first time I run it, it freezes after about 1:30 to 3 minutes, then restarts. I've updated and factory reset it, but that didn't change anything. After every restart, the same FPGA-related entries appear. Has anyone else had a similar problem?

[Martin72]

I’ll take a closer look at that path, because a logic IC is powered by it there.

Bingo....
Now I'm really curious about the surrounding area.
If the 2.5V is used directly for switching, that's a bit silly from a design perspective...
Stay tuned.

Edit:
@Arek:
That almost sounds like a thermal issue.

[goaty]

Hello, I'm having a problem with my VColtcraft. From the first time I run it, it freezes after about 1:30 to 3 minutes, then restarts. I've updated and factory reset it, but that didn't change anything. After every restart, the same FPGA-related entries appear. Has anyone else had a similar problem?
(Attachment Link)
(Attachment Link)

With which version of firmware ? I see two different versions.

Just added a 220u directly to it

Right at the reference output—that's quite a feat.

I think it's ok, as the ref has short circuit proof and limited 18mA output.