Post by Steve Wilson Post by Spehro Pefhany Post by Steve Wilson Post by Spehro Pefhany Post by Joe Gwinn
On Tue, 06 Apr 2021 16:59:07 -0400, Spehro Pefhany
Post by Spehro Pefhany Post by Joe Gwinn
My Agilent 33220A function generator (bought in 2004) died
yesterday, so I need a replacement. (It's too old to bother with
repair, and it's time for a big upgrade anyway.)
I'm looking at a Rigol model DG1062Z. Any opinions and experience?
I have the DG4062 which is similar but 500M s/s sample rate rather
Nice piece of kit. Gorgeous display.
I was looking at that one too. It's very attractive.
But the arbitrary waveform memory is only 16 Kbytes (or samples?),
which won't last long at 500 Msamps/sec. Do they really mean 16 K, or
is the data sheet wrong?
It's correct but you can spread the (up to) 16K points over just about
any period of time up to 10^6s and it will linearly interpolate
between points if desired.
Post by Joe Gwinn
The DG106Z samples at 200 or 25o Msamps/sec, and comes with 8 Mbytes
(or samples?) of memory.
I guess RAM that fast is not cheap.
I have 32 GB of DDR3 running at 1600MHz. Real cheap. Around CAD $196 on
Amazon.ca That's about 196/32e3 = CAD $0.006125 per megabyte, or 163.26
megabytes per dollar.
DDR4 is even faster. 3000MHz, ~CAD $248 on Amazon.
1 GB DDR3 running at 1066 MHz is around CAD $22.
These are all overkill for the application. Slower is cheaper.
1GB DDR at 400MHz is CAD $9.59 on Amazon. I can't find anything smaller.
Newegg is probably cheaper.
PC ram is probably 32 or 64 bits, but I think memory addressing may
return larger chunks. So plain 400 MHz DDR ram may deliver much higher
performance than is needed for this application.
I assume they're using SRAM inside an moderate-sized FPGA, which may
actually be cheaper than buying a discrete SRAM 16K x 16.
16e3*16 = 256,000. They are talking 8 megabytes.
FPGA SRAM has got to be expensive. Kill the RAM and use a smaller FPGA.
Post by Spehro Pefhany
I suspect DDR would add a lot of complexity for not much user benefit.
More memory gives more data points and more waveforms. They can probably
use an existing IC to manage the memory. A complete motherboard is probably
available for under $100, so the memory management portion has got to be
The Raspberry Pi 4 Model B is the latest product in the Raspberry Pi range,
boasting a 64-bit quad core processor running at 1.5GHz, dual-band 2.4GHz
and 5GHz wireless LAN, Bluetooth 5.0/BLE, true Gigabit Ethernet, and PoE
capability via a separate PoE HAT.
The dual-band wireless LAN comes with modular compliance certification,
allowing the board to be designed into end products with significantly
reduced wireless LAN compliance testing, improving both cost and time to
1.5GHz 64-bit quad-core ARM Cortex-A72 CPU (Broadcom 2711)
2GB RAM (LPDDR4 SDRAM) - also available in 1GB and 4GB versions!
On-board wireless LAN - dual-band 802.11 b/g/n/ac
On-board Bluetooth 5.0 HS low-energy (BLE)
2 x USB 2.0 ports
2 x USB 3.0 ports
True Gigabit Ethernet
Extended 40-pin GPIO header
2x micro HDMI, 4k video
4 Pole stereo output and composite video port
MIPI Camera port (CSI)
MIPI Display port (DSI)
microSD format for loading OS & data storage
5V/3A DC via USB type C connector
5V DC via GPIO
Multimedia H.265 decode (4kp60), H.264 decode (1080p60), H.264 encode
(1080p30), OpenGL ES 1.1, 2.0, 3.0 graphics
The extra RAM could be used to add new functions as well as for more data.
They could wipe the competition for not much cost.
Dynamic RAM loses data when powered off, but a cheap flash memory could be
used to save the data and restore it at power on.
They won't be buying RAM from Amazon or Newegg, instead from their own
sources. But the Amazon prices give an indication how cheap memory has
And it is fast!
You would need a fast FIFO that can supply data for about 10
microseconds (IIRC) while the SDRAM refreshes, so about 80K bits. Plus
the DDR interface logic - And a Spartan 3A FPGA with 300k bits of RAM
is not expensives. Not sure if it's fast enough though, prob not.
Still if someone did an open-source RPi-based dual output function
generator using one of those cheap 240 x 320 TFT displays and a PCB
with the FPGA, DAC(s) etc on it, I bet it would be popular even if it
didn't quite measure up to the professional product in one way or
another. There is probably public domain code for the LXI
functionality. Maybe it's usable. There are DDR3 interfaces on
Opencores that claim to be finished.
That's about a $50 Spartan 6 FPGA with 1M bits of RAM, and it is fast
The gorgeous 7" color display, backlit buttons, rugged beautifully
molded case and so on definitely don't come for free.
AFAIK, I have not tested out the output protection, probably it's no
worse (or better) than the Agilent products (which use multiple
paralleled power op-amps on each output - giving off significant heat
at the best of times).
I actually built something sort of like this for a specialized task,
using a Raspberry Pi to handle the offline tasks like loading the
data. Bandwidth was << 25MHz though, so pretty lethargic.