EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: IDEngineer on October 02, 2017, 12:42:57 am
-
We're working on a microcontroller based system that is isolated from "real" ground via an isolated DC-DC module. This is often done in medical systems (though this one isn't medical) for safety reasons. In this case it's done for accuracy reasons.
As a result, we're ordering a small-signal differential probe (Pintek DP-60HS, specifically with a x1 range) to avoid having the scope introduce a connection back to "real" ground. Obviously things like battery operated DMM's are inherently isolated, so no problems there.
The problem is the in-circuit emulator/debugger. This is a Microchip MCU design, and all of Microchip's debuggers are USB hosted. USB grounds back to the host PC, which then grounds back to earth, and bingo - we've violated the isolation barrier. As you might imagine, this introduces all sorts of ground shift problems that make all of our analog measurements completely nuts.
We can't be the only ones who have done MCU development on a ground-isolated circuit and needed to isolate the debugger. Any tips? We're considering an AC isolation transformer for the debugger's host PC, but that really only isolates the hot and neutral leads which means we'd have to manually disconnect the ground connection, leaving the PC's chassis floating :o. I thought about cannibalizing a USB cable to break the shield and have the shield only grounded at the DUT end, but I suspect the Gnd/- signal and power lines will also be hardwired to the PC's ground so that's not likely to yield actual isolation.
Anyone solved this before? Thanks!
-
Use a laptop running on battery for your PC?
-
Use a laptop running on battery for your PC?
Wow, that's actually a very simple, very clever idea. I'm embarrassed I didn't think of it. Probably not suitable for heavy duty work, but for spot debugging that may be just the thing. Thank you!
Any ideas out there for isolating when hosted on a tower PC? It would be nice to have a generic lab-based solution so we don't have to transfer the entire hardware and software environment to a laptop, then find out the laptop battery is discharged, etc.
-
USB isolator? Just got one a few days ago actually, PoUSBiso. Olimex had some too and probs others. Usb full speed only tho.
-
USB isolator? Just got one a few days ago actually, PoUSBiso. Olimex had some too and probs others. Usb full speed only tho.
That's exactly what I'd like to find, but searches so far had revealed nothing. I'll search on those names and see what comes up. Thanks!
-
OK, we're making serious progress now.
Turns out Analog Devices makes a part specifically for this task, the ADuM4160:
(http://www.analog.com/-/media/analog/en/products/image/functional-block-diagrams/adum4160-fbl.png?w=900)
...so all you have to do is find someone that has wrapped some packaging around it and you're in business. Once you know this part number, eBay reveals all sorts of options with all sorts of price tags and countries of origin ;D. I'll probably order up a couple of different ones and see what happens.
Thanks for the tip on the other unit, it led directly to this solution!
-
Looks like the challenge is providing enough isolated DC to the debugger. The Microchip ICD3 requires 300mA for its own operation, but so far the self-contained isolators appear to max out at 200mA. There are a couple that accept an external supplementary power supply, but the isolation from that supply is unclear so far. Research continues.... :popcorn:
-
I'd use a bench psu for it but I imagine not everyone has one around. In that case the olimex one is more flexible, get some 2 prong supplies, 8-15V. Type A plug for the USA i think?
-
You don't have to DIY this. You can just buy it.
UH401-2KV (http://nl.mouser.com/ProductDetail/B+B-SmartWorx/UH401-2KV/?qs=kWcaGGHSxRq3B1L8mA3ivQ%3D%3D)
(http://nl.mouser.com/images/ttelectronics/images/UH401-2KV.jpg) (http://nl.mouser.com/ProductDetail/B+B-SmartWorx/UH401-2KV/?qs=kWcaGGHSxRq3B1L8mA3ivQ%3D%3D)
Or:
https://www.digikey.nl/product-detail/en/segger-microcontroller-systems/8.07.02-USB-ISOLATOR/899-1056-ND/7063173 (https://www.digikey.nl/product-detail/en/segger-microcontroller-systems/8.07.02-USB-ISOLATOR/899-1056-ND/7063173)
Or even lower price:
http://nl.farnell.com/olimex/usb-iso/usb-isolator-1000vdc-for-pc-laptop/dp/1795095 (http://nl.farnell.com/olimex/usb-iso/usb-isolator-1000vdc-for-pc-laptop/dp/1795095)
Or, maybe they sell an isolated programmer. Like ST does, ST-LINK/V2-ISOL, since isolating the slow non-usb stuff is much cheaper and can be done with target power, using SI86xx isolators.
-
[Any ideas out there for isolating when hosted on a tower PC? It would be nice to have a generic lab-based solution so we don't have to transfer the entire hardware and software environment to a laptop, then find out the laptop battery is discharged, etc.
A big UPS ? That's a bit ghetto tho
USB isolators as other mentioned, altho most of them are 12Mbit only, enough for programmer but you wont be streaming any significant data over it so you will have problems with for example trying to use it , and 480Mbit ones will probably be significantly more expensive.
Other option, if you are into DIYing it, is to get a rPi or other small ARM device, install linux and just work with it remotely
-
A big UPS ? That's a bit ghetto tho
Have to disable the freakin' UPS beeper too, since you'd have to unplug it to achieve isolation. ;D
-
USB isolators as other mentioned, altho most of them are 12Mbit only, enough for programmer but you wont be streaming any significant data over it so you will have problems with for example trying to use it , and 480Mbit ones will probably be significantly more expensive.
You can get optical USB 3.0 cables?
-
Update: Linear Technology (yes, normally they're an IC company) sells a packaged USB isolator. They sell the bare PCB version (typical demo board style), but they also sell a fully packaged version that accepts an external DC transformer for higher output current. Here's the link to it on Linear's site:
http://www.linear.com/docs/44955 (http://www.linear.com/docs/44955)
A bit pricier than some other options I've found, but at least with LT you can rely on getting something that works. DigiKey has them in stock. I've ordered one and will report back on how things work out.
-
There are a few threads on USB isolators.
There are a few cheap USB2 (high speed) isolators on the market. I use one with a Saleae and I am very happy with it.
https://www.eevblog.com/forum/chat/high-speed-usb-isolator/msg1197100/#msg1197100 (https://www.eevblog.com/forum/chat/high-speed-usb-isolator/msg1197100/#msg1197100)
-
I've ordered one and will report back on how things work out.
The Linear Tech isolator arrived today and I wasted no time putting it through its paces. We also ordered a dedicated 12VDC 700mA wall wart power supply for it, so there would be no question of sufficient downstream current for the debugger. The wall wart has only hot and neutral prongs so no earth ground reference.
The short story: It definitely works as a USB isolator. The difference is noticeable in our circuit. The error we get is cut by ~70% by using the isolator. (Not sure why it's not 100%, some serious head scratching to be done there but this is a quickie report on the isolator so I'll stick to that.)
The longer story: This is a very nicely packaged piece of test gear. The LED's are especially nice; there are other units on the market but most of them don't have a full array of five LED's telling you what's happening. It's small, takes up very little room on the bench, yet tells you about supplies and current limiting and so forth with a glance. It uses a USB B connector at the input and a USB A connector at the output, so it literally plugs between two standard cables - and they even include an extra USB cable, so it's as simple as unplugging your to-be-isolated device and sticking the isolator+cable between it and the existing USB cable.
When they said "high retention USB connectors" they weren't kidding. I checked a couple of times to confirm I didn't have them misoriented, they were that stiff! Pulling them out is a bear too. Not a problem, just something to be aware of.
The product manual that's included looks exactly like one of Linear's tear-out App Notes that appear in every issue of Electronic Design, right down to the slightly tan paper color.
That's about it. It's really simple, so there's not a lot to report. Frankly, that's a good thing for an isolator - it should transparently sit in your signal path and appear to do nothing, and require nothing.
I'll do a minor "tear down" when I get the chance and post some photos. I suspect it will be very underwhelming.
Bottom line: There are cheaper products that purport to do the same thing. But I have confidence that Linear, who is interested in selling us IC's, will do a good job on what is effectively a demo board showing off their USB isolation chip. Other products are targeting the audiophile market, etc. but this one is specifically intended for Engineers who affect Linear's sales... so I was willing to spend a little more for the implied higher quality. I suspect this will be on the bench for a very long time, so 150 USD and delivery in two days from DigiKey seems very fair.
Recommended!
-
Use a laptop running on battery for your PC?
Or use a laptop power supply that only has two power pins (ie no ground pin). We used to do this when doing data acquisition on large mining machinery (where the presence of a ground pin on the power supply caused ground loop issues and hence noise in the acquired data), and found that many Dell's had such a power supply.
-
Or use a laptop power supply that only has two power pins (ie no ground pin).
Yep, we tried that (it was a good suggestion!). Problem was, everything else connected to the laptop WAS grounded and provided a ground reference. External AC powered monitor (ground through the video cable), other test interfaces that are hosted on the laptop, etc. As I said earlier in the thread, stripping the laptop to its basics works OK for onesie-twosie tests but not for normal development - if only because the screen is too small! ;D
If you're doing regular work on isolated circuits, Linear's device is turning out to be a good choice. It's not perfect isolation - nothing is - but in our tests so far it's pretty darned good. For example, its effective isolation capacitance appears to be in the ~10pF range, quite respectable for something that doesn't even include that criterion on its spec sheet!