1Gs/s Digital scope project. Anyone interested?

[karesz]

I dont know, if I have x-hundreds of pins + their descriptions?
My biggest IC was up to days with 125 pins & PCAD 2001, but if you have more complex as pins up to thousand, you need very much time with conventional methodes, then is a day really too short for that...
I know that Allegro has a tool for the kind of jobs, but I found nothing possibles in in my PCAD descriptions...
K.

 

[farhada]

I have not used PCAD for a long time, but in Orcad, you can create your pin description in EXCEL and then import it into the program, check it out and see if there is such function available for your CAD program.

 

[karesz]

Hi,
These have I meand with Allegro_than Orcad is nowadays Allegro...
K.

 

[manuel1139]

Can we stay on the topic, please?

 

[karesz]

Sorry,
but if I have Pcad & maybe it will come for these project too a Spartan6; than its not so far from the origin of topic...
K.

 

[eziggurat]

In one of my previous company, they used the Atmel ADC 1GS/s and now the 2GS/s. It was linked upto to a Virtex 2 Pro FPGA and all the data manipulation and calculation is performed by the FPGA. It was a 18 layers board. If you have deep pockets, I'm sure it is plausible.

 

[karesz]

Hi Eziggurat,
Do you know some details, data, etcca pls?...
K.

 

[eziggurat]

The Atmel ADC products are now made by www.e2v.com but they are used for RF application. Still similar in data manipulation using a FPGA. You have to use the DDR register in the FPGA to clock the data in. I haven't used a FPGA for a couple of years so my knowledge is not upto scratch as when I was working with them. If you were asking about board layout, I'm afraid that is a bit out of my depth. All I know from working at the company, they used autorouting for some parts of the board (DDR/QDR memory, FPGA, ADC, DAC) with very strict rules. It took the PCB/Circuit designers hours to setup the DRC rules. The software used was Mentor PADS and Hyperlynx. They mostly refer to the application notes and Howard Johnson's book to design their PCBs.

 

[karesz]

Hi eziggurat,
Tnx a lot for your nice info!
Yes, price is surly a problem, but yours projejt is(must be) coming from basic research, or for that developed., becouse the people with 18 layers are surly some luxourious...
I have studied in past year a layout realisation for Spartan 3 project, in FG676 package, inmy opinion it is good to realizable with max 10-12 layers, in our case it becames 8 layers, but we didnt need more as 50% of the pins...
If the pinout of the PGAs good/Iintelligent designed, you has chance to work with pads Nr. of up to ca. 1K(!) with at i.e.12 layers...,but in our digiscope project to, are the must pins not neede, only the much gates, so I believe, that the PCB will be possible not complexer as i.e. 8 layers.
In my opinion is ML(Multi Layer) welcome, but not over all! Nowadays is a board w. 8-layers, not so cheap, but a relative "industrie standard".
K.

 

[FvM]

Starting the hardware discussion with required number of PCB layers is somewhat like put the cart before the horse.
The requirement usually becomes clear when you know about involved chips and number of connection nets. For a fully
wired 484 pin 1mm BGA FPGA, a 6 layer board is usually feasible, it should be for a partly connected larger package as
well. A higher density connection, e.g. using wide busses for RAM modules, most likely needs 8 layers. A today's industry
standard 4 mil/0.1mm structure size is assumed for the design.

I didn't yet hear a specification for the FPGA functionality, except acquiring an either 1 GS/s or 2 GS/s ADC data stream.
So the considerations about a suitable FPGA family aren't exactly precise up to now. Besides the ADC data rate (I understand,
that 8 bit resolution has been assumed so far), particularly the memory specification would be interesting.

 

[eziggurat]

It is a commercial product that has been around since the 2001 and widely used in the surveillance sector. I have also designed a complicated FPGA board schematic a couple of years back and I outsourced to a PCB designer. He also suggests designing it as a 12 layers board due to complex routing based on the ADCs and FPGA used. The contract PCB designer worked for a reputable company which designed high speed processing FPGA board. My FPGA board I designed had 4 high speed transmitters, 2 high speed ADCs and a high end FPGA. This was used for image processing and display technology. Both projects were not coming from a basic research and it was not luxuorious. The layers was needed for the components selected. Another company I know also designed 18 layers and above for their complicated FPGA board but they needed the layers for the components selected. I don't really know what you are talking about when you mention basic and luxurious when you having even seen the final product. For the basic and luxurious board you seem to indicate, I believe that board was chosen to be one of the best 100 products in mid 2000 by notable electronic magazine and the company is still producing the goods. Believe me it is cutting edge technology not basic research.

But for the hobbyist digital scope you won't be designing something similar to these boards. It will depend on what components that you are using.

Hi eziggurat,
Tnx a lot for your nice info!
Yes, price is surly a problem, but yours projejt is(must be) coming from basic research, or for that developed., becouse the people with 18 layers are surly some luxourious...
I have studied in past year a layout realisation for Spartan 3 project, in FG676 package, inmy opinion it is good to realizable with max 10-12 layers, in our case it becames 8 layers, but we didnt need more as 50% of the pins...
If the pinout of the PGAs good/Iintelligent designed, you has chance to work with pads Nr. of up to ca. 1K(!) with at i.e.12 layers...,but in our digiscope project to, are the must pins not neede, only the much gates, so I believe, that the PCB will be possible not complexer as i.e. 8 layers.
In my opinion is ML(Multi Layer) welcome, but not over all! Nowadays is a board w. 8-layers, not so cheap, but a relative "industrie standard".
K.

 

[eziggurat]

I forgot that I answered someone question on how to use the 2GS/s ADC. For your information in how to use this ADC with a FPGA.

 

[karesz]

...I don't really know what you are talking about when you mention basic and luxurious when you having even seen the final product. ... one of the best 100 products in mid 2000 by notable electronic magazine and the company is still producing the goods. Believe me it is cutting edge technology not basic research.

But for the hobbyist digital scope you won't be designing something similar to these boards. It will depend on what components that you are using.

You see, I dont have some problem with them if you says "its a very complex and sensitive PCB-system", I have similars in my praxis too_I can imagin it
Coworkers apllyed on, as Xilinx told it us, Europas biggest FPGA board in 1990-91 (randomly the same ) 18 layers...These PCB was ca. 40x42 cm_so called "Fast Bus" Format, & absolute full with only FPGAs... My highest layer Nr is, up to date max. 8, but mosten mixed with multi Gsps components i.e. 124 smallest charge sensitive preamps, signal shapers + comparators & MUCs self built Optical Hybrids too. COB, THT & SMD realized, also I have some idea over the problematics of cuch projects. My most problem was in those projects the lot of different supply voltages + GND systems to place/route properly Their are as first version problemless installed and in function over 7 years, up to deinstallation of the full measurements system. My Optoelectronic Board was at Electronica 2000 demonstrated from PCB producer "the as absolute cutting edge Star-Fex technology" of year...Has had 8 layers, 100um lines COB & much impedance lines (burried too) & 4layers as Kapton...
OK, I make finish.
I know, all thes isnt for a hobbyst scope, but its is not my Idea, I sayd only; I can learn some too with the project & I like to make it "up to end of ADC", & I propos in two versions ; 1 for 100Msps and 2 in 1Gsps for possigle full service of interested peoples.
I know, you can by newest Digiscopes for ca. $300.- at 40-100Msps & some better from China i.e. for ca. 600.--......
Thes means me, that it is a real target only as techniqual interest success.
K.

 

[eziggurat]

Sorry Karesz, I am only giving you my experience working with a high speed FPGA board and how the boards were designed by my ex-company. Like I said I am no expert but my ex-colleagues were experts in their own field.

 

[karesz]

Hi Eziggurat,
Nothing is a problem_ and I wish best thanks to say for other aspects too! So, i.e. was the Atmel as ADC source for me unknowed...
We are/was fixed (with coworkers) for ADI & TI ADCs...
K.

 

[ANTOTRONIC]

Take a look at my HSD-15030 scope:

**broken link removed**

(use google translate for now, english version will be available soon)

This is ready to use board that contains:

- ADC08200 ADC
- Xilinx FPGA + Flash configurator
- National Instruments' analog front end with 30 MHz analog bandwidth and 10 mV/div sensitivity. Bandwidth can be exceeded to 70 MHz by reducing sensitivity.
- high impedance input suitable for typical oscilloscope probe
- USB interface, optically separated when using external power supply
- can be also powered from USB interface

Example application samples at 150 MS/s. You can use this board as hardware for your own experiments. I can sell it width schematic diagram and software.

This is not 1GS/s 250 MHz scope, but belive me, designing 1GS/s sort of device from scratch is NOT THAT SIMPLE. It's good to practice with simplier devices.

Best regards,
ANTOTRONIC

info@antotronic.com
**broken link removed**

My other (free) designs:
http://www.antoniak.ep.com.pl/

 

[karesz]

Tchetch Anton,
The ADUC812 is a really nice component in your "Fast logic probe", & with ca. $15 is to pay too...
Greetings!
Karol