Needing more than a spark test?

[WobblyHand]

They probably don't think anyone would be so foolish as to use a USB connection to provide power and signal transfer to/from a 16-bit ADC .

Really, the only reason for having a USB connection is to transfer data (and in the case of developing Teensy code) to program the processor.

Interestingly, a Teensy is programmed over a 480Mbps link (if available on USB3) You would have to modify a bunch of low level code to get it to work at 12Mbps. 12Mbps isn't fast enough for a 1 MHz conversion rate and a 16 bit converter, even if one was silly enough to try...

Well, that link/reference went over my head. Sorry I needed to ask for an explanation.

Back to the regularly scheduled programming.

After complaining to Digikey that they can't possibly ship me my product unless they actually give it to the common carrier, they have created a replacement order and have actually transferred the new package to FedEx. FedEx says they have the package and it seems to be moving through the system. Maybe a Friday delivery, we'll see. Ordered the ADC, the buffer, a bunch of op amps, and some SMD stuff.

Mark, do you think this ADC board is salvageable with some rework? Can you post the Gerbers or artwork? Is this a 4 layer stack? I can see the split, clear through the board, but obviously its tough to work out some of the vias. Outer layers with routing?

Maybe with some creative surgery and or drilling the vias, the capacitor grounding can be fixed? Just looking for an assessment from you.

 

[graham-xrf]

Something like this might help resolve some of the issues we're dealing with.

Yes Yes - good thought!

If the a 5V came from a supply (wall wart, whatever), got regulated clean to something above 4.75V, and powered the analogs, with DVDD (pin 19) using the 20Ω and decouplers, then a 3V3 taken from a Teensy can power the interface. A somewhat cleaner 3V3 can even be had regulated with LDO, starting out from a computer with 5V, so long as it does not share GND. It will work perfectly OK returning from OGND (pin 17).
That one seems unavailable, but there must be more like it around.

This need not be too awkward. All we need is a 3V3 supply to work the data transfer interface, and not try to run currents in the analog instrumentation circuits GND.

It is not too critical, and need not be particularly "clean", although it helps if anything that gets close to an ADC is not rough as hell.

 

[homebrewed]

Interestingly, a Teensy is programmed over a 480Mbps link (if available on USB3) You would have to modify a bunch of low level code to get it to work at 12Mbps. 12Mbps isn't fast enough for a 1 MHz conversion rate and a 16 bit converter, even if one was silly enough to try...

Well, that link/reference went over my head. Sorry I needed to ask for an explanation.

Back to the regularly scheduled programming.

After complaining to Digikey that they can't possibly ship me my product unless they actually give it to the common carrier, they have created a replacement order and have actually transferred the new package to FedEx. FedEx says they have the package and it seems to be moving through the system. Maybe a Friday delivery, we'll see. Ordered the ADC, the buffer, a bunch of op amps, and some SMD stuff.

Mark, do you think this ADC board is salvageable with some rework? Can you post the Gerbers or artwork? Is this a 4 layer stack? I can see the split, clear through the board, but obviously its tough to work out some of the vias. Outer layers with routing?

Maybe with some creative surgery and or drilling the vias, the capacitor grounding can be fixed? Just looking for an assessment from you.

While it would be most convenient to program the Teensy and then immediately use it to test the code, it should be possible to switch over to the USB isolator when actually using the Teensy as the XRF pulse processor. In that mode the USB emulates a slow serial interface.

I have attached a PDF showing the top and ground plane layers; and also the schematic to help identify the relevant connections. The most challenging will be the bypass capacitor connected between OVDD and OGND/DGND because it is on the bottom side of the board underneath the AD7667. It might be easiest to just lift the pin(s), although making connections to them wouldn't be a piece of cake.

The LDO's ground connection is a via inside the ground pad. Relatively easy to break that connection.

 

[homebrewed]

Yes Yes - good thought!

If the a 5V came from a supply (wall wart, whatever), got regulated clean to something above 4.75V, and powered the analogs, with DVDD (pin 19) using the 20Ω and decouplers, then a 3V3 taken from a Teensy can power the interface. A somewhat cleaner 3V3 can even be had regulated with LDO, starting out from a computer with 5V, so long as it does not share GND. It will work perfectly OK returning from OGND (pin 17).
That one seems unavailable, but there must be more like it around.

This need not be too awkward. All we need is a 3V3 supply to work the data transfer interface, and not try to run currents in the analog instrumentation circuits GND.

It is not too critical, and need not be particularly "clean", although it helps if anything that gets close to an ADC is not rough as hell.

Hmm. I just ordered two. They actually seem to be relatively common, at least judging by the number of hits I got on Amazon when I looked for "USB isolator". Their main application appears to be aimed at protecting your computer from stuff on the other end of the USB cable.

 

[graham-xrf]

There is a ton of those isolators on Amazon
Duh - your post crossed with my reply, and we were both on Amazon!

 

[graham-xrf]

Hmm. I just ordered two. They actually seem to be relatively common, at least judging by the number of hits I got on Amazon when I looked for "USB isolator". Their main application appears to be aimed at protecting your computer from stuff on the other end of the USB cable.

They are cheap enough perhaps. I get itchy about every twenty bucks here and there.

Yes, their designed purpose is to protect the computer from stuff on the other end of the USB, We want the XRF stuff on the other end of the USB to be protected from the computer (while stealing some energy from it)!

 

[homebrewed]

Can you post the Gerbers or artwork? Is this a 4 layer stack?

The ordering method I employed didn't require generating a Gerber, but it was easy enough to do: attached.

Yes, it's a 4-layer stack.

Something to add to the "design revision list" is a series damping resistor on the CNV\ input. Maybe the same on BYTESWITCH, since both of them need to work correctly in the tens of nS timeframe. If not, easy enough to put a zero-ohm resistor in its/their place.

 

[WobblyHand]

The ordering method I employed didn't require generating a Gerber, but it was easy enough to do: attached.

Yes, it's a 4-layer stack.

Something to add to the "design revision list" is a series damping resistor on the CNV\ input. Maybe the same on BYTESWITCH, since both of them need to work correctly in the tens of nS timeframe. If not, easy enough to put a zero-ohm resistor in its/their place.

Installing ~22 ohm resistors at the source is a good thing in general. Kills ringing.

 

[homebrewed]

They are cheap enough perhaps. I get itchy about every twenty bucks here and there.

Yes, their designed purpose is to protect the computer from stuff on the other end of the USB, We want the XRF stuff on the other end of the USB to be protected from the computer (while stealing some energy from it)!

Well I went with the "cadillac" model. I found some bare-bones versions for about half that

Fleshing out the "revision list", I'm open to any/all suggestions. My current thoughts:

1. Completely separate DGND and OGND planes.
2. Separate +5 input for OVDD LDO, with a jumper to permit using the Teensy's 3.3V supply
3. Series damping resistors for all the ADC's control lines.
Question: what about the data output lines? It would be best to have damping resistors on the receiving end but that would require modding the Teensy. Possible but probably not easy for the average hobbyist to accomplish.
4. Test points to ease debugging with a DVM and oscilloscope.
5. Change the physical connections so the ADC can be plugged into the Teensy. This will greatly shorten the signal lines, which should help regarding signal integrity.

Regarding #5, the data pins will have to be different if plugging into a T4.0 vs. a T4.1. I don't like jumpers but they may be the most expedient way to customize the ADC board. The other option: don't design it to accommodate a 4.0. A 4.1 isn't much more expensive; and the ADC driver code is a lot cleaner.

Anything else?

 

[WobblyHand]

Well I went with the "cadillac" model. I found some bare-bones versions for about half that

Fleshing out the "revision list", I'm open to any/all suggestions. My current thoughts:

1. Completely separate DGND and OGND planes.
2. Separate +5 input for OVDD LDO, with a jumper to permit using the Teensy's 3.3V supply
3. Series damping resistors for all the ADC's control lines.
Question: what about the data output lines? It would be best to have damping resistors on the receiving end but that would require modding the Teensy. Possible but probably not easy for the average hobbyist to accomplish.
4. Test points to ease debugging with a DVM and oscilloscope.
5. Change the physical connections so the ADC can be plugged into the Teensy. This will greatly shorten the signal lines, which should help regarding signal integrity.

Regarding #5, the data pins will have to be different if plugging into a T4.0 vs. a T4.1. I don't like jumpers but they may be the most expedient way to customize the ADC board. The other option: don't design it to accommodate a 4.0. A 4.1 isn't much more expensive; and the ADC driver code is a lot cleaner.

Anything else?

Damping series resistors should go at the source, near the ADC. Want to damp the ringing at the source, not transmit it along a pseudo-transmission line, spraying rubbish everywhere. Kill EMI and conducted problems at the source. Parallel terminations (we don't need any) go at the destination. In this case, nearly 100% of the time, the ADC is the source. I think you can control the effective drive strength for the Teensy outputs, if needed.

Yes, for test points, can't have enough of them.

I'd recommend against modification of the Teensy - they are already densely packed in some areas and would be tough to modify except by very skilled people. I'm not sure I could do it, and I do have most of the equipment. My opinion, keep it as a commodity item and add complexity to the ADC board, if required.

Re:#5. Although I have both now, it would be a lot more convenient for me to use the 4.1. It also avoids the byte swapping stuff, which is one less area to worry about. KISS.

What is the total supply current for the ADC and buffer? The Teensy could supply some of the current, but I don't know if it is wise to, since it is on a digital board.

Jumpers are both good and bad. In a way, I like 0 ohm resistors better. You can use them to bridge tracks, and they offer a lower profile. They are easy to place and remove (relatively) if at least 0603 and larger. Jumpers are somewhat more convenient, but are not good for trying to cram a board into a tight place.