FEDEVEL
Platform forum

slime , 01-14-2025, 12:30 AM

Hello, I wanted to try and complete my first hardware project that’s been shelved for a very long time. It’s for a Nixie Clock, and I need help finding out a way to go about creating a level translation from 3.3v from my ESP32-S3-WROOM-1 microcontroller to a high-voltage shift register 12v.

I’m not sure if Im reading into this datasheet correctly, but it looks like the chip im using has some strict requirements and needs a maximum of 1 μA for highs and -1 μA for lows. It also allows clock lines to run at 8MHz so I would need something that keeps up with that too.

I see a lot of people who decide to drive them with a BJT inverter circuit (as shown below) but I’m really confused because whenever I try to simulate it in LTspice it seems like the clock speed is way too fast and the low logic current is nowhere near a microamp.

Would anyone here have any recommendations on how to construct such a driving circuit that meets this criteria? I feel like I may be overcomplicating things or forgetting something and I wanted to ask around.

HV Shift Register Chip Datasheet:
https://www.microchip.com/en-us/product/hv5530

Reference Schematic from Another Project:
https://github.com/GadgetReboot/
HV5222_Breakout/blob/ymain/Sch_HV5222_Breakout.pdf

(This example breakout board uses an arduino hence the 5V vin in the spice simulation)

QDrives , 01-14-2025, 08:49 PM

That the shift register can handle 8MHz, does not mean you have to **do** 8MHz.
You can go a lot slower.

Robert Feranec , 01-15-2025, 06:15 AM

exactly as @QDrives says .... do you really need 8MHz? Is that what you are going to send out from ESP32? If this is used to control the clock segments, you should be able to run it at much lower speed

slime , 01-16-2025, 01:06 AM

Ah forgive me, you’re both right. I was originally using the HV5222 which is from the same family and they had 8MHz as a recommended condition so I may have gotten the two mixed up. I can definitely pull that down to something as low as the Hz range if its just to change the digits.

Do you also think that I should still be concerned about what I highlighted for the criteria for IH/IL? Specifically with the reversed current for IL.

I don’t believe the behavior of this switching circuit consistently holds a negative value for a low-logic level (There are only negative dips in the transition from high to low as shown above).

QDrives , 01-16-2025, 04:57 PM

That is in nA.
Simulations have these kind of 'problems'. Noting to worry about.

slime , 01-21-2025, 07:25 PM

ESP32 Nixie Tube Project Review

slime , 01-21-2025, 07:28 PM

Back again, and I’m at the part where I have a full project architecture laid out.

Currently trying to create a power budget for it and I wanted to know if I’m going about it correctly.

So far, I’ve been building a table of individual components with an estimate of their power consumption.

Since, my design has multiple power sources, according to the conservation of power law wouldn’t I just need to find the total consumption for each sub-rail and then from that, select each power source based on the requirements of their respective loads?

(e.g. Treat Σ 3.3V * Rail Current Draw, Σ 5V * Rail Current Draw, and Σ 12V * Rail Current Draw separately to draft up requirements)

Wanted to see if creating this sort of High-Voltage/Low-Voltage mix power network from a USB-C VBUS was reasonable. Excuse the newbish question, just want to be very certain is all.

QDrives , 01-21-2025, 08:53 PM

For a linear regulator (e.g. LDO) the input current = output current + Iq (quiescent current)
For switching regulators (e.g. fly-back) input power = output power / efficiency.

slime , 01-21-2025, 11:32 PM

Thanks for looking over again, I really appreciate it.

I’m aware that the regulators/LDO have special equations regarding their power I think I maybe was more trying to ask if the way I calculated the consumption for each power source is correct (P=VI for each component per rail)?

Apologies if I worded it weird.

Like should I just base the pout / iout values in the equations you just mentioned based on the all the current/power consumed for the specific rails each device powers?

QDrives , 01-21-2025, 11:35 PM

As I tried to explain, linear (LDO) is **current** whereas switching (fly-back) is **power**.

QDrives , 01-21-2025, 11:38 PM

So if you are using linear, it does not matter what the power is, the current consumption * input voltage is the power required.
For the 170V and 12V you can use power and efficiency.

slime , 01-21-2025, 11:49 PM

Okay I think I seem to understand a little bit clearer.

slime , 01-21-2025, 11:49 PM

If power isn’t relevant for selecting an LDO but current is, for 3.3V I should take the current demands into consideration and power for everything else when searching for components then?

QDrives , 01-21-2025, 11:53 PM

If you consume 340mA on 3.3V, you consume 340mA on 5V.
If you consume 0.1W at 12V, it is 0.1W on 5V.
Neither of these take Iq and efficiency in consideration, nor does it have any other consumers on the voltages.

QDrives , 01-21-2025, 11:55 PM

340mA @ 3.3V = 1122mW, 340mA @5V = 1700mW.
0.1W @ 12V = 8.3mA, 0.1W @ 5V = 20mA

slime , 01-22-2025, 09:47 PM

Right, but that just suggests that I would need to add the contributions that a power source causes on both sides of the rail to everything else consuming in my budget to keep track of total consumption for each rail/the circuit no?

As you mentioned before, the LDO creates a contribution to the total current drawn on both the 5V rail and 3.3V rail.

Which should be equal to:

Load current on 3.3V rail + (IQ from LDO)

Also shouldn’t the switching regulator(s) downstream from 170V->12V->5V since I have the two cascaded.

So if I were go back to your example, if I wanted 0.1W at the 170V rail, and the datasheet says I have an efficiency of 90% corresponding to whatever load I need on the 170V rail, then I’d need 0.111W coming in from the 12V line into the Flyback.

And then for that 0.111W output from the switching regulator I would have to repeat the calculation to find the power I need from the 5V rail.

QDrives , 01-22-2025, 09:57 PM

Correct.

slime , 01-22-2025, 10:04 PM

Alright thanks once more! Now back to planning again :)