66319B Firmware image?, datalogging, and other interesting info

[pigrew]

I'd like to update the firmware of my (newly purchased) 66319B to A3.x.x in order to use its datalog mode.

Is there a copy of that ROM image available online?

If not, I'll buy it through Keysight. ($22+shipping+tax vs $4 for the AT27C2048 PROM)

Any clue if it would be a good idea to run the ROM on a 66309B? (Though, I doubt that it would add any new features).

Thanks!

[technogeeky]

I'd like to update the firmware of my (newly purchased) 66319B to A3.x.x in order to use its datalog mode.

Is there a copy of that ROM image available online?

If not, I'll buy it through Keysight. ($22+shipping+tax vs $4 for the AT27C2048 PROM)

Any clue if it would be a good idea to run the ROM on a 66309B? (Though, I doubt that it would add any new features).

Thanks!

Any update on this? I am now interested too.

[pigrew]

I ordered the ROM. I had trouble with them setting up my purchasing account (by default they only allow businesses email addresses to buy parts), and I ordered over the phone, with a $20 surcharge.

They ended up sending me what seemed to be unprogrammed ROMs. I've sent them back, and am waiting for them to confirm there is an issue with them.

They said there should be a part number sticker on them, but there was not. The ROMs they sent were packaged in 2014.

I've taken apart the supply so many times that I'm worried about damaging all the connectors....

Buying the firmware has turned into a frustrating and slightly expensive experience.

[technogeeky]

I ordered the ROM. I had trouble with them setting up my purchasing account (by default they only allow businesses email addresses to buy parts), and I ordered over the phone, with a $20 surcharge.

They ended up sending me what seemed to be unprogrammed ROMs. I've sent them back, and am waiting for them to confirm there is an issue with them.

They said there should be a part number sticker on them, but there was not. The ROMs they sent were packaged in 2014.

I've taken apart the supply so many times that I'm worried about damaging all the connectors....

Buying the firmware has turned into a frustrating and slightly expensive experience.

I am considering buying one of these units, but I am a little bit torn. The price for this unit is pretty much comparable to the price for a used GPD-3033S from Instek, and the specifications are actually a lot more similar than I expected. The primary differences seem to be that the 66319B has the fast transient handling, current sinking, resistance set mode, and a lower current readback resolution. Negatives for the 66319B are that I will have to get a GPIB controller, it's significantly deeper, VFDs seem to go bad over time, and I'm sure it's tremendously more complicated in design and to repair (and I don't have and can't afford the software, so I'd have to roll my own). On the other hand, the 3033S has two identical 30V channels, already has banana jacks on the front, displays the voltage and current setpoints for both channels at the same time, and it supports USB. And negatives for the 3033S include it's at least $50 more, I find it hard to believe the spec (e.g. ripple/noise) is really as good as the HP unit.

Do you think you can find time to take some ripple/noise oscilloscope screenshots loaded and unloaded, and perhaps an example of exactly how fast the turn-on and turn-off time is?

By the way, what use case did you buy this unit for? What perks did you anticipate enjoying that fall outside of your primary use case? (for example: It should serve as a nice milliohm calibrator for multimeters).

[pigrew]

I am considering buying one of these units, but I am a little bit torn. The price for this unit is pretty much comparable to the price for a used GPD-3033S from Instek, and the specifications are actually a lot more similar than I expected. The primary differences seem to be that the 66319B has the fast transient handling, current sinking, resistance set mode, and a lower current readback resolution. Negatives for the 66319B are that I will have to get a GPIB controller, it's significantly deeper, VFDs seem to go bad over time, and I'm sure it's tremendously more complicated in design and to repair (and I don't have and can't afford the software, so I'd have to roll my own). On the other hand, the 3033S has two identical 30V channels, already has banana jacks on the front, displays the voltage and current setpoints for both channels at the same time, and it supports USB. And negatives for the 3033S include it's at least $50 more, I find it hard to believe the spec (e.g. ripple/noise) is really as good as the HP unit.

Do you think you can find time to take some ripple/noise oscilloscope screenshots loaded and unloaded, and perhaps an example of exactly how fast the turn-on and turn-off time is?

By the way, what use case did you buy this unit for? What perks did you anticipate enjoying that fall outside of your primary use case? (for example: It should serve as a nice milliohm calibrator for multimeters).

I'm using it to measure the power draw of wireless radios. At least, I think that's what I'm doing with it. I'm dealing with design using a BLE transceiver board. I don't expect to use the battery resistance mode, as our batteries are more like 20 ohms series resistance.

I've also used it as a general bench supply, for example powering microcontroller boards/GPSDO/whatever. Repair is somewhat impractical due to a lack of schematics. I've never used channel 2, nor have I used the DVM. Note that the DVM voltage can't float too far away from the main output's negative.

I've already written some Windows software to control it fairly well. It's posted on Github. I wrote it initially for my 66309D.

I don't have time to do proper noise measurements right now; I expect that I'd have to build some amplifier circuits, etc, to do it justice.

Big advantages of the HP supply:

• Record current vs time, as if the supply as if it were an oscilloscope, reading 4096 samples at 66 kSa/s (one channel only). With the datalogging mode (if I ever get the updated ROM), I should be able to do nice plots of current consumption over time. With the Instek supply, you'd have to use an external multimeter. It seems to be a sampling measurement, instead of an integrating measurement, so it will often not record fast pulses.
• Low-current measurements: the HP supply has a 20 mA range
• Protection-mode where it turns off the supply when the current limit is met, etc.

If you want fast turn-on/turn off, you'd need the MOSFET switch option. The supplies have mechanical relays by default. By fast, they mean that when the load changes suddenly (like when the RF amplifier turns on), the output voltage will hold steady.

I'd prefer a Keithley 2400/2602, but this supply meets 98% of my needs.

[pigrew]

Spoiler alert: I got the ROM (finally).

I've not put it through all its paces, but it seems to be working.

I even recorded the ROM installation procedure and transcribed it into closed captions:

.

Filming it took MUCH longer than I expected. It was definitely a learning experience.

Anyway, I'll post the ROM at some point soon, I know a few other people have also expressed interest in it. I need to figure out the exact mechanics of triggering data logging over GPIB, also, and I'll integrate that into my control software once I get a chance.

[nidlaX]

I am considering buying one of these units, but I am a little bit torn. The price for this unit is pretty much comparable to the price for a used GPD-3033S from Instek, and the specifications are actually a lot more similar than I expected.

If you're not doing precision measurements or battery simulation, there's no point to buying a mobile comm DC supply. I have a 66309D and a Siglent SPD-3303D, and I can tell you that the Siglent is the goto for general purpose use, especially because it's so quiet and the controls are very convenient.

[technogeeky]

I am considering buying one of these units, but I am a little bit torn. The price for this unit is pretty much comparable to the price for a used GPD-3033S from Instek, and the specifications are actually a lot more similar than I expected.

If you're not doing precision measurements or battery simulation, there's no point to buying a mobile comm DC supply. I have a 66309D and a Siglent SPD-3303D, and I can tell you that the Siglent is the goto for general purpose use, especially because it's so quiet and the controls are very convenient.

For one, noise just isn't a problem for me. My lab is noisy and there's nothing I can do about it. I do get the thing about easier controls. Ultimately the 66319 was too expensive ($160+), and the next cheapest unit available was the Keithley 2403A for $125 shipped. The GPD-3033S would have been about $160 too.

I'm actually quite happy with the Keithley. It's quite a bit less noisy (electrically) and it's incredibly fast. It's even worse in terms of front panel controls, and I will have to do GPIB. But I would need to be scripting anyway to do the things I want to do (SMU-like parameter sweeps of devices) and battery simulation. The primary downside is I don't have output impedance control. But I can live with that.

On the other hand, the Keithley 2403A has a virtually unlimited amount of room in the front in order to add/modify things; namely adding one (or two or three) banana jack sets on the front and/or adding a different kind of display. There's even room to try and make a crappy knock-off of the modern Keithley touch-screen interfaces. I think that's probably excessive.

[technogeeky]

Spoiler alert: I got the ROM (finally).

I've not put it through all its paces, but it seems to be working.

I even recorded the ROM installation procedure and transcribed it into closed captions:

.

Filming it took MUCH longer than I expected. It was definitely a learning experience.

Anyway, I'll post the ROM at some point soon, I know a few other people have also expressed interest in it. I need to figure out the exact mechanics of triggering data logging over GPIB, also, and I'll integrate that into my control software once I get a chance.

Hey that's an impressive start. Keep making youtube videos!

[pigrew]

On the other hand, the Keithley 2403A has a virtually unlimited amount of room in the front in order to add/modify things; namely adding one (or two or three) banana jack sets on the front and/or adding a different kind of display. There's even room to try and make a crappy knock-off of the modern Keithley touch-screen interfaces. I think that's probably excessive.

Yes, the Keithley is a good choice too. I'm tempted to buy a second 2403A at the current eBay price, just because.

I've not confirmed it (yet), but one difference is that I believe when measuring current, the Keithley unit integrates the current over a length of time, whereas the HP supply only does a spot measurement. Thus, the Keithley 2304A may be able measure power more accurately (it wouldn't miss fast spikes). However, it doesn't have the all-powerful datalogging mode that lets the instrument sample the current over minutes (or hours). Then again, perhaps 663xxA are actually integrating ADCs without a sample-and-hold? I'm not too sure.

Hey that's an impressive start. Keep making youtube videos!

Thanks for the support. I was gearing this video as more of a detailed repair video. Sometimes I find myself watching videos where the mechanical actions are skipped over, and it leaves me wondering about good soldering techniques or how to fit wrenches into tight spaces.

I plan to continue to film repairs as they come up. Perhaps I'll demo the MCU I've put together for a GPSDO in a future video. I think that I'd better invest in a better microphone (I was using my Olympus E-M10's internal microphone this time).

[pigrew]

I've (mostly) figured out the Datalogging commands. They've (mostly) been integrated into my control program for Windows.

(I'm not sure if I'll investigate CCDF/histogram mode.)

First, set your data format to REAL (either swapped or NORM)

Code: [Select]

FORMat REAL; BORDer [NORMal|SWAPPED]
Then, configure datalogging, read it back (to get true window length), and start it up:

Code: [Select]

CONF:DLOG [CURR|VOLT|DVM],[0.02|1.0|3.0],[DC|ACDC],[window length in secs],1024,IMM
CONF:DLOG?
INIT:NAME DLOG
TRIG:ACQ
*ESR? // Check for errors

Then, reading data by periodically calling:

Code: [Select]

FETCh:ARRay:DLOG?
It'll return a standard binary-block of 32-bit floats. Note that it will be terminated with a newline, which must be read, or else you'll get synchronization errors. The standard read binblock command doesn't read the newline.

The first three elements is the length of elements in the buffer, duplicated three times. Then each element follows (mean,min,max).

So, for example if the min is -1mA, mean is 0 mA, and max is 1 mA, you could get [2,2,2,0,-1,1,0,-1,1]. When you read the array, it'll return all available data, up to its buffer size of 1024.

In the case of a buffer overflow (you didn't readback data quickly enough), it'll return [-1,-1,-1] and clear the buffer. Sometimes it'll recover, but usually you'll need to abort and restart the measurement.

To stop the measurement:

Code: [Select]

ABORT
The power supply's clock seems to be within 40 ppm of my laptop's clock. For my software, I'm not sure what the best way to handle inaccuracies in the power supply's timebase. Currently, the software assumes that its timebase is accurate. I was thinking about trying to fudge the resulting times based on the computer's clock, but I'm not sure if that's much better (since my laptop isn't running NTP).

My computer can read back without buffer overfills when the window length is set to about 5ms. The maximum seems to be 1.00001 seconds (any value above this is ignored).

You can only measure on parameter (curr, volt, or dvm) at a time. The official Agilent software gets around this by periodically stopping datalogging, reading the other parameters, and restarting datalogging.