Author Topic: EBAZ4205 (Zynq-7000 based development board) (Read 26747 times)

thinkfat · « **on:** December 18, 2020, 03:51:29 pm »

Hi,

there's a large number of these boards currently popping up on ebay and aliexpress, word is they're control boards from decommissioned bitcoin miners.

Anyone around here already familiar with those? Artix-7 FPGA with a dual ARM Cortex-A9 cluster strapped on its back.

Khaveer · « **Reply #1 on:** December 19, 2020, 07:37:15 pm »

I've ordered one of those about a week ago. I'm still waiting for it to be shipped. There's a github repository with schematics and additional information: https://github.com/xjtuecho/EBAZ4205

RoGeorge · « **Reply #2 on:** December 19, 2020, 09:42:44 pm »

https://hackaday.com/2020/11/18/hacking-the-fpga-control-board-from-a-bitcoin-miner/
https://hackaday.com/2020/12/10/a-xilinx-zynq-linux-fpga-board-for-under-20-the-windfall-of-decommissioned-crypto-mining/

Dave · « **Reply #3 on:** December 21, 2020, 04:25:27 am »

I've ordered mine in early November, been playing with it for around 2 weeks now. I ordered it just out of curiosity, because I couldn't believe anyone would be selling ZYNQs with all of the required surrounding circuitry for that little. It didn't make sense, but I bought it anyway. Only when I found out later, that this was a repurposed board from an ASIC Bitcoin miner, did everything suddenly click into place.

There are plenty of resources for this board online, but most of it is in Chinese (Google Translate has been rather helpful). Would be rather nice if we had more people working with it here.

Neekeetos · « **Reply #4 on:** December 22, 2020, 06:19:11 am »

For this money it is very good. Even better if you solder in some parts like PL clock... Im currently making a frequency meter out of this. Having raw time resolution of less than 10ps and ability to capture every edge of input signal I can store this data to DDR (up to 256MB) and process/analyze using dual core arm and then pass processed data via ethernet, and all this on a single board...

CJay · « **Reply #5 on:** December 22, 2020, 11:34:04 am »

Been looking at these too, for the price, they look excellent value so I've ordered a couple.

What toolchain are people using?

FlyingDutch · « **Reply #6 on:** December 22, 2020, 02:17:55 pm »

Most likely Xilinx Vivado (together with Xilinx SDK for programming ARM hard CPU). Petalinux can also be taken into account.

Regards

FenTiger · « **Reply #7 on:** December 22, 2020, 04:25:26 pm »

Quote from: Neekeetos on December 22, 2020, 06:19:11 am

raw time resolution of less than 10ps

How do you manage this?

Neekeetos · « **Reply #8 on:** December 22, 2020, 04:42:13 pm »

Quote from: FenTiger on December 22, 2020, 04:25:26 pm

How do you manage this?

сarry chain delay line with latches. For this board ( slowest speed grade artix ) , single carry4 gives 70ps total delay. This is around 17ps delay per tap, if you are using 4 of them. After calibration rms error is ~ half of the tap delay.

thinkfat · « **Reply #9 on:** December 22, 2020, 08:09:15 pm »

You compare the signal to taps with increasing delay, to find the taps between which the edge is. The tap order tells you the temporal location of the edge.

FenTiger · « **Reply #10 on:** December 23, 2020, 03:52:18 am »

Interesting.

I have a breadboard/prototype of a "continuous timestamping" frequency counter on my bench which uses an XC7Z010 FPGA talking to TDC7200 ICs to actually sample the input edges.

It's somewhat disappointing to think that I could have achieved ~5x better resolution by omitting the TDC7200s and just doing it all in the FPGA. Doh!

Perhaps I'll give this a go when I get back to the project. Thanks!

Neekeetos · « **Reply #11 on:** December 23, 2020, 05:49:30 am »

Quote from: thinkfat on December 22, 2020, 08:09:15 pm

You compare the signal to taps with increasing delay, to find the taps between which the edge is.

It is rather complicated way of describing things. Delay line contents is captured every clock cycle of fpga, in my case it's 400Mhz. This creates say 100+ points
of input signal , each of different delay . After calibration this is just equivalent of capture of signal with sampling rate 1/tau = 1/70ps/4 ~ 57GHz

this is the basic structure

delay line with decoder after implementation, marked yellow and green

capture of 125MHz external signal , slowly drifting about own fpga clock..

Quote from: FenTiger on December 23, 2020, 03:52:18 am

I have a breadboard/prototype of a "continuous timestamping" frequency counter on my bench which uses an XC7Z010 FPGA talking to TDC7200 ICs to actually sample the input edges.

I could only guess that with TDC7200 you are having hard time counting actual edges. It has large dead time because of GRO . I once tried to implement GRO type TDC , but this was on spartan6 and result was rather noisy.

FenTiger · « **Reply #12 on:** December 23, 2020, 06:55:23 am »

I'm not sure what you mean by GRO - can you explain? - but I'm not having any problems counting edges with it. The design works fine - it's just limited to the 55ps precision of the TDC7200. I haven't observed any "dead time" with it.

Neekeetos · « **Reply #13 on:** December 23, 2020, 07:03:30 am »

Quote from: FenTiger on December 23, 2020, 06:55:23 am

I'm not sure what you mean by GRO - can you explain? - but I'm not having any problems counting edges with it. The design works fine - it's just limited to the 55ps precision of the TDC7200. I haven't observed any "dead time" with it.

GRO = Gated Ring Oscillator. The method itself based on free running oscillator which is started with input start signal. After that, phase of input signal relative to ref clock is deduced via counting of ring oscillator cycles to phase match condition. Thus, while counting gro cycles you can not register a new event = dead time.

FenTiger · « **Reply #14 on:** December 23, 2020, 07:14:03 am »

OK. I see what you mean. Yes, that's how the TDC7200 works.

I wouldn't call that "dead time" in my counter because it doesn't delay the next measurement; the TDC7200 only counts for a couple of hundred nanoseconds, then there's some time spent pulling the measurement out of it via SPI, but none of this is "dead time" as such because the coarse counter is still running and it's still possible to measure the delta between the next edge and the reference clock without losing any time.

I don't try to measure every edge. I can measure about 100,000 edges per second. With a 20MHz input signal that still lets a good few slip by.

Anyway this is probably getting way off topic for this thread

mawildoer · « **Reply #15 on:** January 07, 2021, 09:13:00 pm »

Just ordered mine with the intention of building a basic ethernet-based oscilloscope/DAQ/something which resembles the red pitaya. Let's bet I'll be back with questions 😁

CJay · « **Reply #16 on:** January 08, 2021, 08:48:49 am »

Mine just arrived, I specified with SD card slot and the seller sent it with the slots installed, the resistor moved to boot from card and the headers for JTAG and UART populated.

Weirdly didn't send it with D24 installed so it wouldn't power up.

I stole the diode from one of the IO ports temporarily so it powers up and works perfectly but to a very limited Busybox.

Next step is to download Vivado (It's *huge*) then work out some way to switch between NAND or SD and begin to experiment.

Impressive little board and really good price.

thinkfat · « **Reply #17 on:** January 08, 2021, 11:14:05 am »

Mine should be in the mailbox any day now. I also ordered with SD card slot installed. Let's see what I got when it's finally here.

CJay · « **Reply #18 on:** January 08, 2021, 12:42:21 pm »

Quote from: mawildoer on January 07, 2021, 09:13:00 pm

Just ordered mine with the intention of building a basic ethernet-based oscilloscope/DAQ/something which resembles the red pitaya. Let's bet I'll be back with questions 😁

Oh now *that's* an interesting idea, it hadn't occurred to me it's the same chip.

Might have to order another couple

Khaveer · « **Reply #19 on:** January 10, 2021, 02:15:17 pm »

My board also recently arrived. I ordered the board without the SD card slot but they included one anyway along with .1 inch headers for JTAG and UART. For some reason I also got one male to male jumper cable all packed in an ESD shielding bag. The board itself was packed in one of those pink bubble-wrap bags.

There was a ton of dust all ofer the board but after a quick IPA bath and some scrubbing it was good as new. Same as CJay I had to steal a diode from one of the other connectors.

The Molex power connector is the same as those used in PC power supplies, so I was able to steal one from a scrap unit. From my initial experiments it would appear that the board can be powered using voltages lower than the specified 12V. I was able to go down to 5V and it was still running fine however I didn't yet test it under full load.

Also I had to solder an adapter to connect my Digilent HS-3 JTAG adapter due to a different pin pitch.

RoGeorge · « **Reply #20 on:** January 10, 2021, 05:45:08 pm »

Very tempting price to order yet another toy, but I already have a box full of all kinds of unused devboards.

Can anybody list a few more projects good for this board, so to lower the buying guilt, please?

mon2 · « **Reply #21 on:** January 10, 2021, 06:22:47 pm »

Quote from: RoGeorge on January 10, 2021, 05:45:08 pm

Very tempting price to order yet another toy, but I already have a box full of all kinds of unused devboards.

Can anybody list a few more projects good for this board, so to lower the buying guilt, please?

Bud · « **Reply #22 on:** January 10, 2021, 06:31:53 pm »

"How to build a picture frame" - Home Depot

tchicago · « **Reply #23 on:** January 10, 2021, 09:53:47 pm »

Quote from: Bud on January 10, 2021, 06:31:53 pm

"How to build a picture frame" - Home Depot

Electronic LCD picture frame that is *actually* usable and useful? Hmm, interesting project and FPGA can find a good use in there

agehall · « **Reply #24 on:** January 11, 2021, 09:52:58 am »

Damn you EEVBlog - I had promised myself that I wouldn't buy new toys in January. Now there are 5 boards on the way from China..

I'm not sure what I'll do with them, just like the other 3 Zynq boards I have here already, but I really like the Zynq concept and I have a few ideas I'd like to try at some point in time.


EEVblog Main Site	EEVblog on Youtube	EEVblog on Twitter	EEVblog on Facebook	EEVblog on Odysee

Author Topic: EBAZ4205 (Zynq-7000 based development board) (Read 26747 times)

Share me