HP 34401a DMM with leaking segments

[bitseeker]

Probably you will sell a truck of them. Please consider a group order.

Doing logistics of group order and delivery is not worth it to me. Board designs as well as everything else will be open so anyone can do it.

Yep. I might look into organizing a group buy for this. It may be OK in the form of kits (i.e., DIY soldering). Getting it completely manufactured might be too much of a project. We can revisit after qu1ck is done.

[qu1ck]

Heh. I thought of completely replacing the front panel PCB too. It is doable but will need even more firmware development (I have button codes but I don't currently emit anything onto data lines). Also manufacturing that pcb will be quite costly.
As floobydust mentioned it has lots of cutouts and would need precision machining, as well as tight tolerance on thickness to fit snugly and gold plating for button pads and it's pretty big and... yeah it's gonna cost ya. Nowhere near the $2 for 10 bargain that I can get for tiny 2 layer mcu board.

[bitseeker]

Yeah, it's another level for sure.

[floobydust]

... handling of the buttons would still need to be added, right? I assume that's what's meant by "finish the replica PCB."

At first I thought of making a small daughter board to replace only the NEC uPD7527 but not enough space.
Then I figured just replace the entire front panel PCB, keeping the VFD. New tubes are plentiful from China.
I started with the mechanical, including the keypad etc. Then the H/W and F/W design is how I usually go. Getting a PCB made to verify how it fits, it's on the list.

I don't want to compromise the instrument by having a noisy, high current MCU running off the +/-18VDC rails which run to the A/D converter. Designing for conducted EMI and keeping it quiet to keep life rosy at 6-1/2 digits is important.
I think an underclocked Bluepill would be OK, or Mega328PB- something with two SPI ports and under 12mA. The original uPD7527 is 3mA, and 12mA for 87C51.

To bring the multimeter up to date, adding wireless is tempting. WiFi needs a lot of TX current, so power would have to come off the 5VAC filament winding. ESP8266 is max. 170mA chirps but the second SPI port would have to be bit-banged I think. ESP32 seems like overkill? But only $5.

I haven't really nailed down what I'm doing.
This is the keypad drawing, one key needs vias moved but the slots and VFD and bracket crap I have also figured out.

I can try add qu1ck's OLED and see how that fits. The huge 5.5" OLED was too big.

[qu1ck]

You have to bit bang SPI either way. In 34401a protocol FP notifies the master of readiness to receive next char or button press event by pulling MISO line low and waiting for the clock. I don't think that kind of shenanigans are normally supported in hardware SPI on most MCUs, but I may be wrong.

Thanks for the drawing, it will come handy.

[qu1ck]

Did first pass on the board. I would appreciate feedback. I've done much more complex boards before but still, this is not my expertise.

Some notes: height, mounting holes and 16 pin header match the display board so that you solder it directly on top (maybe with a pin header as a strut, which you can cut off later). I noticed now that it should be 33.5mm high, not 35, that's an easy fix.
Data lines are short and nice, 5v data inputs are fed through 10k resistors to protect stm32 just in case. Those pins are supposed to be 5v tolerant.
UART header and boot state solder bridge are only there to make flashing bootloader easier. After that you will always use usb bootloader.
Power is supposed to be fed from unregulated U553 input on the dmm, that's the line that is feeding earth referenced 5v. I don't think that line is sensitive at all, correct me if I'm wrong. That power input will also go to display Vcc.
Voltage reg is placed on a big polygon, but I kinda eyeballed the power dissipation. Need to check the math on that.

Also ignore the pin headers on the render, they won't be there in end product.

[HighVoltage]

What PCB software are you using?

[qu1ck]

Kicad 5 rc3.

Sent from my moto g(6) using Tapatalk

[qu1ck]

Added eeprom, cleaned up and annotated pcb, fixed height.

It's going to the fab! Even if I screwed something up, I'll be only out $10.

[floobydust]

I wasn't sure about the physical location, 12V power source etc.

The OLED modules I've looked at have a different pinout. Not sure what OLED part you are using, I'm thinking of the 3.12" offerings. The modules with on-board boost converter are 5V powered with 3.3V I/O to match the SSD1322, using around 150mA at 5V. I think 12V input will heat up the vreg too much.

10k series resistor on MISO is too high, consider R517 215R and R506 5k62 5V pullup on the main ASIC and CPU, I don't think you'll get a logic 0 there.

I use a series resistor between 5V outputs and 3.3V inputs, "5V tolerant" sometimes means the IC has diodes to +3.3V

[qu1ck]

floobydust, thanks for reply. Display I'm using is this one, it has different pinout from one on your pic: https://www.buydisplay.com/download/manual/ER-OLEDM032-1_Series_Datasheet.pdf
It is powered the same way but I plan to rip out the boost converter from the display and feed 12v directly where boost output is to minimize switching noise. 3v3 from voltage reg will also go in on the standard pin to power ssd1322.
Re MISO resistor, what does it matter what the pullup is? All the pins on that 10k RN are configured as inputs. My board does not output any signals, only listens on what is going on the CPU->FP data lines. It's the front panel micro that will be fighting against the MISO pullup. (Or rather the comparator they use to shift voltage levels, but the point stands).

I calculated heat dissipation on voltage reg and assuming there is 0.5inch^2 of copper attached to it I can dissipate about 0.5 watt before it goes past ~60C. For 8-9 volt drop that's about 55-60mA. I am hoping that  stm32 + ssd1322 - high voltage consumption of ssd1322 will fit into that. Some quick searching did not give any specific numbers for either of these chips, I guess it varies a lot depending on used peripherals and display update frequency.
As a backup I can always tap from regulated 5v of the DMM instead.

[qu1ck]

Poor chinese bastards have no weekends I already got a tracking number yesterday.

While I'm still waiting for the boards I though I'll show Kicad bom generating script I wrote a year ago. Adapted it to Kicad 5 as a plugin now, here is result (it will be deleted after a day):
http://ibom-showcase.bitballoon.com/

It doesn't support arbitrary polygon shapes, rounded rectangle pads, other things, but it is still useful. I used it before on a much more complex board, highlighting similar components was very handy for hand soldering hundreds of smd components. Let me know what you think, if others like it too I might make an effort to support pcb format more fully and I'll open source it.

[Zucca]

It's going to the fab!

Sorry for asking, which one?

[Kosmic]

While I'm still waiting for the boards I though I'll show Kicad bom generating script I wrote a year ago. Adapted it to Kicad 5 as a plugin now, here is result (it will be deleted after a day):
http://ibom-showcase.bitballoon.com/

Really cool, good job! 

[qu1ck]

It's going to the fab!

Sorry for asking, which one?

Allpcb
Their quality is good from my experience, although silkscreen is not highest resolution print. But nothing beats $10 prototype boards with express fabrication and free dhl shipping.

Sent from my moto g(6) using Tapatalk

[bitseeker]

While I'm still waiting for the boards I though I'll show Kicad bom generating script I wrote a year ago. Adapted it to Kicad 5 as a plugin now, here is result (it will be deleted after a day):
http://ibom-showcase.bitballoon.com/

Really cool, good job! 

That is a very cool tool!

[Zucca]

Really cool, good job! 

That is a very cool tool!

Nice one, qu1ck is now in my buddies list.

[qu1ck]

Well this is a hilarious fuckup (images clickable):



Do you see it?

How about now?



And yes, I double checked my gerber files, they are obviously correct
I'll try to scrape the solder mask off, maybe this batch is salvageable still.

How did they even manage to do this? Well, back to the fab with ya.

[bitseeker]

Wow, not just one pad, but two of them. Maybe a bug in their software? Both cases were similar in that the trace to the pad continued straight through. However, there's a trace on the lower left that went through the pad to a via and that didn't get solder masked. Odd.

[Zucca]

I'll try to scrape the solder mask off

Yes, it will be not pretty but it should work.

[qu1ck]

Yeah, only ground plane pads got affected and only 2 of the 4. 
Fret not, some elbow grease and good microscope to the rescue:







Also had some fun with recording video. I just might record the whole assembly process too.

https://youtu.be/ZoB_NMqq5YA

[bitseeker]

Looks good, qu1ck. If you record the assembly, I'll watch.

[qu1ck]

I'm waiting on couple parts to arrive before I start the build. But in the meantime I published the Kicad BOM plugin I showed earlier.

Github

Better than earlier demo

Started a thread for it in KiCad section https://www.eevblog.com/forum/kicad/interactive-html-bom-plugin/

[hillflyer]

How about displaying the 1 or 2 extra digits that are available via GPIB or RS232?

[Samogon]

How about displaying the 1 or 2 extra digits that are available via GPIB or RS232?

Since he is getting data given to front panel i think it is not possible.