Electronics > Projects, Designs, and Technical Stuff

A Universal Programmer

<< < (2/2)

colejohnson66:
The DAC I’m using cam work with any voltages with a delta of 20V. I figured that going with a delta of 10V and then quadrupling would be easier than generating 10V and -10V supply lines, but I’m already going to be generating 22V and -22V just for the opamps. Anyways, I guess I could change it to -3.25V to 6.75V which would give me the -13V to 27V Ian suggested. The opamps work with any voltage swing of 44V from my cursory reading of the data sheet.

My design is already complex enough for my first real design and first board I’d be ordering, so I’d like to keep it as simple as possible. I could add a -40V rail/bias that could be switched to, but then I’d need to add another demuxer, and that would just complicate the logic of the MUX_SEL line (lines if I add more demuxers). In addition, like I said, this won’t be a true universal programmer like the UniSite, but it will be able to do more than most.

Re: DIN41612: I saw that the UniSite uses this to connect the module to the system. I’m currently debating between that and a card edge connector like PCIe. For example, a PCIe 4x slot has 64 pins (48 pins with 16 left over for whatever) and a female connector costs just shy of $3 on DigiKey whereas DIN41612 requires a ~$2.50 receptacle (female) and a $3 plug (male). Now, I’ve never had a PCB manufactured before, so I don’t know how much card edge connectors cost, but from my understanding, it’s just exposed copper, no?

Also, I do plan to make this expandable, but maybe not right away. Maybe for a second revision? If I do do that, maybe a PCIe 16x for 164 connectors (160 pins with 20 driver boards with 4 leftover). If I go with DIN41612, it looks like the biggest standard size is 96 pins. I could use multiple connectors (2x64?), but that would just double the interconnect costs. I’m also wary of using actual pins as they could bend. A card edge connector doesn’t get bent pins. Anyways, that’s just for the interconnect to the actual DIP module.

For the interconnect on the pin driver boards, I was thinking of using standard 0.1" (2.54mm) 2x20 headers/sockets. This is partly evident in the schematic. From a quick Google search I did a while ago, they can handle about an amp per pin. With a maximum pin current of 100 mA, that wouldn’t be a problem. I could use a DIN41612 (48 pin) connector here too, but 0.1" headers/sockets are dirt cheap compared to DIN41612.

colejohnson66:
Update:

I have changed the MOSFETS to 5V drive voltage ones (BSS1138L (N) and BSS84 (P)). I have also added some op amp buffering through some ZXTD4591E6 NPN/PNP BJTs to increase the output current. And lastly, I changed the DAC/op amp output from -20V~20V to -13V~27V.

FrankBuss:
I think you have the MOSFETs wrong. The n-channel MOSFET in your circuit diagram is always conducting. You can test it yourself with LtSpice:



or even better, you should try it on a breadboard. And read my previous message about driving MOSFETs. You can't do this with 5V, if you want to switch higher voltages than 5V. See here for more information:
https://electronics.stackexchange.com/questions/188745/high-side-driver-and-low-side-driver

colejohnson66:
Thank you. I’ve been avoiding breadboarding because my DAC and opamp are only available in surface mount, but I guess I could just use a SMT to DIP adapter in the meantime. As for the MOSFETs, I’ve done some digging, and it appears the TL866 (based on a reverse engineered (I’m assuming) schematic) uses BJTs for the voltage switching on the pin. I understand MOSFETs are more suited for power applications, but what would be the downsides of using BJTs with, I assume, half an amp (at most) running into the pin?

And lastly, I’ve used LTSpice before, but it feels very clumsy to use. Is there anything better, or should I just put up with it?

FrankBuss:
I don't know much about analog electronics, but here are some answers:

https://www.quora.com/Why-is-MOSFET-better-than-BJT

I have a box with all sorts of SMT to DIP adapters, they are cheap on eBay.

LTSpice is the industry standard for simulating circuits. I think if you use it more often, it doesn't feel clumsy. And much faster than testing something on the breadboard. If it doesn't work in LTSpice, it usually doesn't work in reality either. But if it works in LTSpice, you still have to test it in reality, but chances are better that it will work :)

Navigation

[0] Message Index

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod