DC to DC conversion help needed for USB device

Hi all,

I have designed and prototyped a basic working model USB Oscilloscope, based on a PIC32. I am now looking at the analogue input section, and am having difficulties presenting a constant 1 Megaohm impedance as the CM Voltage is 2.5 rather than 0. I have a couple of ideas:

1. Make the input differential rather than referenced to ground.

2. DC to DC conversion: +5V to +/-5V - This looks good as it can have the added benefit of isolation from the PC.

I know there are USB scopes available that run solely from the USB power, like Picoscope etc, so there must be a satisfactory solution - does anyone have any idea how commercial USB powered scopes do it?

Also if I am using a DC to DC converter, what is the best way to do this? Bearing in mind I want to keep the size and cost down, and also provide a very clean supply for the ADC and OpAmps. Would something like this be any good?
https://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=5442516

Any help would be appreciated.

Oli

There was a very similar topic not long ago. I recommended the DC-DC converter solution. If you want gain switching which maintains the input impedance but you are not doing it relative to ground, want to measure DC signals and want to use a single positive supply it gets very messy and complicated.

The DC-DC converter you linked to is unregulated. That probably isn't a bad thing - I would suggest you create +/-3.3V for the circuitry using linear regulators. That will help to get rid of the rubbish on the power supply lines.

Keith.

Hi Keith,

I think the similar post was also mine

I decided to go down the DC to DC route you suggested as it seems far better than the alternatives, but as there are so many choices of IC, Module, etc I thought I'd ask for help on the subject. For the MCU 3.3V supply I have already ordered a LDO regulator so that should be fine.
The main thing is I am worried about the noise from the switching on the +/- 5V analogue supply - I ordered a couple of inductors and largish caps to filter the ripple, but I obviously can't use a regulator as there is no voltage to spare. I would like to know the best way to proceed, are you suggesting I have the analogue circuitry using +/-3.3V instead of 5V? If I can I would rather keep it at +/- 5V as otherwise I have to change the opamps and ADC. Would basic filtering after the converter be sufficient?

Oli

My concern with using the 5V, is I don't know what the regulation is like, nor how noisy it will be. If you want to use it then you need some inductance to try to clean it up. Maybe 4.7uH for the high frequencies. You may need a 2 stage filter with a larger inductor as well. You would also need to pick a regulated DC-DC converter.

If you are running you MCU on 3.3V, I would do the same with everything else.

Keith

Hi

You can use a DC2DC converter and an LDO at the DC2DC out put
This method will decrease your noise level by about 50~70db

remember to design the DC2DC op voltage some 300 to 500 MV above the LDO minimum input

All the best

Bobi

If you don't think that the chip is old, i suggest you to use MC34063A from Motorola, and with minimum components it can do your Job. I only hope that the current demand for -5V won't be much.
You might get hold of the datasheet where in application examples , including possible pcb artwork were given.

bobcat1 said:

Hi

You can use a DC2DC converter and an LDO at the DC2DC out put
This method will decrease your noise level by about 50~70db

remember to design the DC2DC op voltage some 300 to 500 MV above the LDO minimum input

All the best

Bobi

Click to expand...

The MAX710 does the job for you in one chip - regulating the switcher 100mV above the output voltage then linearly regulating. I am not sure they do a negative version though. I have used the MAX710 in battery powered instrumentation and it is excellent.

Keith

Thanks for all the replies.

I thought about running the whole lot at 3.3V, and I suppose for the sake of a bit of time replacing the parts, it makes the most sense.

I could also find a DC to DC that has an output above 5V.

I'll have a look at both options, and the chips that have been mentioned. I need an isolated converter, so it cant be transformerless like the MAX710, I should have mentioned this, sorry.

On another note, what does anyone think about the necessity of isolating the data lines also? I know a couple of scopes do this but to me, it seems the most important thing is for the scope to not share ground with the PC, and the chances of a fault reaching all the way to the data lines are almost non existent. The main thing is whether I can still market it as isolated.

I did look at isolating them, but there is no easy, cheap way to do this (there is a chip from AD but it's expensive and hard to get hold of) so IMHO it is not worth it. I'd be interested in peoples views on this though.

for such purposes (very low noise) I made USB dc/dc ZVS flyback.