Author Topic: FAST A>D conversion  (Read 5776 times)

0 Members and 1 Guest are viewing this topic.

Offline stjTopic starter

  • Super Contributor
  • ***
  • Posts: 2195
  • Country: gb
FAST A>D conversion
« on: April 04, 2019, 11:41:40 am »
what's the fastest way to do 6bit A-D at TTL levels??
can i manage atleast 20M-samples?
 

Offline jonroger

  • Regular Contributor
  • *
  • Posts: 72
  • Country: us
Re: FAST A>D conversion
« Reply #1 on: April 04, 2019, 01:38:44 pm »
A parallel output ADC such as the AD9280?
I am available for custom hardware/firmware development.
 

Offline OM222O

  • Frequent Contributor
  • **
  • Posts: 768
  • Country: gb
Re: FAST A>D conversion
« Reply #2 on: April 04, 2019, 01:53:21 pm »
can I ask why you want 6 bit ad TTL levels? that's quite course ... about 0.1V which is not really useful for any analog circuit.
if you're trying to build a crude logic analyzer, your best bet is voltage compactors, not an ADC. if you desperately want an ADC however, fastest ADCs are the flash topology which is available in the low resolutions that you are asking (i.e: 6 bit).
 

Offline stjTopic starter

  • Super Contributor
  • ***
  • Posts: 2195
  • Country: gb
Re: FAST A>D conversion
« Reply #3 on: April 04, 2019, 03:06:27 pm »
i want to convert a digital(supposedly) 5v signal into 6bit digital so i can read the logic state in .1v resolution.

in short, i want to make a logic probe that does not just indicate "high" or "low" but attempts to judge the signal level to detect failing i.c. outputs.
 

Offline OwO

  • Super Contributor
  • ***
  • Posts: 1250
  • Country: cn
  • RF Engineer.
Re: FAST A>D conversion
« Reply #4 on: April 04, 2019, 03:14:31 pm »
AD9200 is only $1 and gets you 10 bits @ 20MSPS.
Email: OwOwOwOwO123@outlook.com
 

Offline OM222O

  • Frequent Contributor
  • **
  • Posts: 768
  • Country: gb
Re: FAST A>D conversion
« Reply #5 on: April 04, 2019, 03:24:01 pm »
i want to convert a digital(supposedly) 5v signal into 6bit digital so i can read the logic state in .1v resolution.

in short, i want to make a logic probe that does not just indicate "high" or "low" but attempts to judge the signal level to detect failing i.c. outputs.

it all really comes down to how much bandwidth you want and how cheap you want the project to be. as I mentioned before, flash ADCs are the fastest but they come in low resolution and are more expensive, but can have a few GSPS bandwidth.a few MSPS can be done with any generic ADC tbh. I strongly recommend you go on mouser / digikey, search for ADCs and then play around with the filters. you can select by topology, cost, sample rate, resolution, etc to find a good balance. one thing I suggest is to stick with I2C or SPI because if you can't find an existing library, other interfaces are generally more painful to work with.
 

Offline stjTopic starter

  • Super Contributor
  • ***
  • Posts: 2195
  • Country: gb
Re: FAST A>D conversion
« Reply #6 on: April 04, 2019, 05:14:44 pm »
sounds like a good start. thanks.
 

Offline jonroger

  • Regular Contributor
  • *
  • Posts: 72
  • Country: us
Re: FAST A>D conversion
« Reply #7 on: April 05, 2019, 04:08:37 am »
I'd review the math on getting 20Msps across an I2C or SPI bus.
I am available for custom hardware/firmware development.
 

Offline stjTopic starter

  • Super Contributor
  • ***
  • Posts: 2195
  • Country: gb
Re: FAST A>D conversion
« Reply #8 on: April 05, 2019, 10:47:20 am »
i was thinking about a parallel bus for that reason,
i²c is out, SPI i dont know, but it would probably need a pretty large microcontroller.
 

Offline OM222O

  • Frequent Contributor
  • **
  • Posts: 768
  • Country: gb
Re: FAST A>D conversion
« Reply #9 on: April 05, 2019, 12:45:08 pm »
yeah my bad ... I've never needed that high of a sample rate in my projects so SPI and I2C were fine  :-DD parallel output would be your best bet in this case.
 

Offline IDEngineer

  • Super Contributor
  • ***
  • Posts: 1950
  • Country: us
Re: FAST A>D conversion
« Reply #10 on: April 05, 2019, 01:46:01 pm »
You guys beat me to the post about serial vs. parallel interfacing on high data rate A/D's. I've never had to work with really high data rate converters, but this always seemed to me like the biggest challenge. You're generating 10's or 100's of MB or more, continuously, in real time. That's dedicated discrete hardware at least, or perhaps an FPGA. Then if the data stream is truly continuous you have to dual-port the memory, or multi-bank it, or something. Sounds like a fun challenge but definitely a non-trivial one.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Re: FAST A>D conversion
« Reply #11 on: April 05, 2019, 01:51:20 pm »
i want to convert a digital(supposedly) 5v signal into 6bit digital so i can read the logic state in .1v resolution.

Logic analyzer which can read logic state as voltage is... every modern oscilloscope. Just saying.
 
The following users thanked this post: tooki

Offline David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Re: FAST A>D conversion
« Reply #12 on: April 05, 2019, 09:38:31 pm »
There are lots of cheap TTL compatible parallel output 8 and 10 bit converters which become 6 bits if you ignore 2 or 4 least significant bits.
 

Offline splin

  • Frequent Contributor
  • **
  • Posts: 999
  • Country: gb
Re: FAST A>D conversion
« Reply #13 on: April 05, 2019, 10:20:40 pm »
The LPC4370 is a 204MHz triple core ARM Cortex processor with an M4 and two M0 cores with an 80MHz 12 bit ADC  - overkill perhaps but they are only $5.5 (10 off). BGA package only unfortunately. Digikey don't even sell 3.5MSPS 12 bit ADCs for that price!

If you only want one you can buy an LPC-Link 2 development board for 18 Euros from Embedded Artists. They used to sell a Labtool board which plugs onto an LPC-Link 2 board to make a 2 channel 60MSPS digital scope and/or 11 channel 100MSPS logic analyzer. The software is a bit limited but the firmware and software are available so you could save yourself a lot of work if you went this route.

You might want to have a look at this thread which uses the 20MSPS 8 bit ADC1175 for < $2:

https://www.eevblog.com/forum/projects/4-channel-adc-10-mhz-8-bit-design/

« Last Edit: April 05, 2019, 10:24:25 pm by splin »
 

Offline stjTopic starter

  • Super Contributor
  • ***
  • Posts: 2195
  • Country: gb
Re: FAST A>D conversion
« Reply #14 on: April 06, 2019, 12:07:20 am »
this may be a good time to add that i'm trying to keep it small.
i was thinking of just streaming the adc to a latching buffer, then a 5bit binary>decimal convertor and finally an led bargraph, using the 6th bit to change the led colour to indicate 0-2.5 or 2.5-5v
 

Offline NorthGuy

  • Super Contributor
  • ***
  • Posts: 3286
  • Country: ca
Re: FAST A>D conversion
« Reply #15 on: April 06, 2019, 01:18:19 am »
this may be a good time to add that i'm trying to keep it small.
i was thinking of just streaming the adc to a latching buffer, then a 5bit binary>decimal convertor and finally an led bargraph, using the 6th bit to change the led colour to indicate 0-2.5 or 2.5-5v

Are you going to update the bargraph at 20 MHz?
 

Offline Someone

  • Super Contributor
  • ***
  • Posts: 5154
  • Country: au
    • send complaints here
Re: FAST A>D conversion
« Reply #16 on: April 06, 2019, 01:29:46 am »
this may be a good time to add that i'm trying to keep it small.
i was thinking of just streaming the adc to a latching buffer, then a 5bit binary>decimal convertor and finally an led bargraph, using the 6th bit to change the led colour to indicate 0-2.5 or 2.5-5v
Are you going to update the bargraph at 20 MHz?
Charlieplexed to keep the pincount low....
 

Offline splin

  • Frequent Contributor
  • **
  • Posts: 999
  • Country: gb
Re: FAST A>D conversion
« Reply #17 on: April 06, 2019, 02:33:42 am »
this may be a good time to add that i'm trying to keep it small.

Most of the ADCs seem to be at least 24 pin TSSOPs; the 42MSPS 8 bit ADC08351 is cheap ($1.45 @ 1K) and is available in WQFN-24 but is a 3V part.

Quote
i was thinking of just streaming the adc to a latching buffer, then a 5bit binary>decimal convertor and finally an led bargraph, using the 6th bit to change the led colour to indicate 0-2.5 or 2.5-5v

You also need an oscillator and possibly an i/p buffer depending on the source impedance. How about the STM32L412K8? It has 2 ADCs which can run up to 8.88MSPS in 6 bit mode. They can be interleaved to give you 17.76MSPS. It is < $2 which is probably cheaper than the ADC solution as you wouldn't need an oscillator or binary to dec convertor. You would likely need a buffer as the maximum source resistance is 220 ohms at this speed. It is available in a 5mm x 5mm UFQFPN - 32.

It is a 3.3V part only but it would give you a lot of flexibility for handling glitch detection - the ADCs have 3 'analog watchdog' comparators which can generate interrupts when user specified thresholds are crossed. You could for example extend the LED on time when a glitch is detected to improve its visibility (a 50ns LED pulse would be very hard to see without overdriving the LED - but you could use a capacitor to do that easily enough I guess)..

You might also be able to detect oscillation and indicate its frequency on another bargraph. You'd not have much processing time for each sample as you'd only get 4.5 clocks per sample @ 80MHz clock but you could use the DSP instructions to process 4 x 8 bit samples per instruction. I'm not sure if it can be configured to DMA the ADC data to a GPIO port directly to drive the LEDs - that would remove a lot of software overhead. If not you could use a second DMA to transfer the samples from memory to a GPIO.
 

Offline OwO

  • Super Contributor
  • ***
  • Posts: 1250
  • Country: cn
  • RF Engineer.
Re: FAST A>D conversion
« Reply #18 on: April 06, 2019, 04:50:03 am »
AD9200 is only $1 and gets you 10 bits @ 20MSPS.

Where to buy? I usually ignore Taobao and other Chinese non-authorized sellers.

Digikey has it listed at $8, LCSC at $9, and even ADI official store at $3. The cheapest authorized reseller is Quest, at $2.7@1k pcs, or Chip One Stop Japan (an Arrow company), at $2.8@1 pcs.
It's around $1 on aliexpress and $0.5 on taobao. It probably is a Chinese replacement but I have used over 300pcs from taobao from 3 different sellers and every single one performed to spec over the specified environment and I haven't had any fail after year. I don't care about getting genuine ADI parts, only parts that perform to spec and don't fail. Anyone should be allowed to make a AD9200 and sell it as AD9200 if it adheres to the datasheet specs.
Email: OwOwOwOwO123@outlook.com
 

Offline stjTopic starter

  • Super Contributor
  • ***
  • Posts: 2195
  • Country: gb
Re: FAST A>D conversion
« Reply #19 on: April 06, 2019, 11:20:30 am »
to answer some questions,
no mutiplexing, just 30 leds wired direct to the binary>decimal convertor.
update rate of the leds would be the rate the data under the probe changes state - that's why the latch is between the dac and binary>decimal decoder.
now 20MHz is kind of worse case - most signals would be 4MHz or less - i picked 20 because it's kind of the max you would see a 74ALS series being run.
i know they can go higher but it's very rare.
 

Offline NorthGuy

  • Super Contributor
  • ***
  • Posts: 3286
  • Country: ca
Re: FAST A>D conversion
« Reply #20 on: April 06, 2019, 12:45:30 pm »
now 20MHz is kind of worse case - most signals would be 4MHz or less

But if your output is LED, the best you can possibly see is 20 Hz or so.
 

Online joeqsmith

  • Super Contributor
  • ***
  • Posts: 12277
  • Country: us
Re: FAST A>D conversion
« Reply #21 on: April 06, 2019, 12:58:53 pm »
to answer some questions,
no mutiplexing, just 30 leds wired direct to the binary>decimal convertor.
update rate of the leds would be the rate the data under the probe changes state - that's why the latch is between the dac and binary>decimal decoder.
now 20MHz is kind of worse case - most signals would be 4MHz or less - i picked 20 because it's kind of the max you would see a 74ALS series being run.
i know they can go higher but it's very rare.

So you want to make something like the following but with a higher BW?

Offline stjTopic starter

  • Super Contributor
  • ***
  • Posts: 2195
  • Country: gb
Re: FAST A>D conversion
« Reply #22 on: April 06, 2019, 05:41:02 pm »
now 20MHz is kind of worse case - most signals would be 4MHz or less

But if your output is LED, the best you can possibly see is 20 Hz or so.

i can pause the latch on the buffer to hold the sample - or maybe feed it with a 2-3Hz clock
 

Offline stjTopic starter

  • Super Contributor
  • ***
  • Posts: 2195
  • Country: gb
Re: FAST A>D conversion
« Reply #23 on: April 06, 2019, 05:44:55 pm »
to answer some questions,
no mutiplexing, just 30 leds wired direct to the binary>decimal convertor.
update rate of the leds would be the rate the data under the probe changes state - that's why the latch is between the dac and binary>decimal decoder.
now 20MHz is kind of worse case - most signals would be 4MHz or less - i picked 20 because it's kind of the max you would see a 74ALS series being run.
i know they can go higher but it's very rare.

So you want to make something like the following but with a higher BW?


no, that's a 2D display, i'm after a single row of 30led's
your on the right track though, imagine a VU meter with a 20MHz+ frontend.
 

Offline nctnico

  • Super Contributor
  • ***
  • Posts: 28429
  • Country: nl
    • NCT Developments
Re: FAST A>D conversion
« Reply #24 on: April 06, 2019, 06:28:32 pm »
to answer some questions,
no mutiplexing, just 30 leds wired direct to the binary>decimal convertor.
update rate of the leds would be the rate the data under the probe changes state - that's why the latch is between the dac and binary>decimal decoder.
now 20MHz is kind of worse case - most signals would be 4MHz or less - i picked 20 because it's kind of the max you would see a 74ALS series being run.
i know they can go higher but it's very rare.

So you want to make something like the following but with a higher BW?

That is neat. I made exactly the same thing... over 30 years ago. Only mine had a 16 wide * 8 height display.
« Last Edit: April 06, 2019, 06:30:11 pm by nctnico »
There are small lies, big lies and then there is what is on the screen of your oscilloscope.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf