mini h-bridge for driving VFD filament very warm to the touch

[Zero999]

Good to know about the spikes - thank you! Should we add a capacitor or inductor to the input then to try to mitigate this?

So does this mean that if the datasheet says 5.6Vac typical, that I can do 5.6V output with this circuit without worry? Is paying attention to the current draw the more critical factor?

You mentioned that we can't really increase the resistor values that much because it needs a certain amount to drive the base, which is why I was wondering about the capacitors. I'll try it as is and see how it works out.

What do you think about surface mount 600mA BJT PNP transistors (for example, MMBT4403)? Anything special that I should look out for? I'm trying to minimize footprint on the board so would prefer to stick away from through hole transistors.

Thanks again for all your help!!

An inductor in series with the positive connection to emitters of Q3 & Q4 (try 10µH) will limit the current spikes. Current limiting can also be added to reduce the surges.

The transistors aren't critical, as long as they can pass the required current.

I don't know much about driving VFDs, as it's not something I've done before, but it's just a filament, so the frequency won't be important. I designed this circuit to make a loud, low voltage buzzer.

The resistor values can probably be increased a bit, with no problems.

[Ian.M]

OTOH if your MCU can generate PWM synchronous to your multiplexing frequency it might well be a better choice (and probably more compact) to use a MOSFET H-bridge IC driven by 50% duty cycle PWM from your MCU at the multiplexing frequency or a multiple of it.  You can fine-trim the filament RMS voltage by adding deadtime to the PWM.

Must Read: Noritake's A Guide to Fundamental VFD Operation, or PDF here

Another compact approach to VFD filament drive is to use a dedicated VFD filament driver. Unfortunately the LM9022 is NLA, but a substitute is readily available https://www.eevblog.com/forum/projects/vfd-filament-power-using-lm4871-replacement-for-obsolete-lm9022/

[Zero999]

The BD6211F-E2 is an example of an H-bridge IC which will work at 5V.
http://www.farnell.com/datasheets/2096555.pdf?_ga=2.128000716.1303208652.1596274850-1107606318.1553553996&_gac=1.247072304.1596274987.EAIaIQobChMI-6Sjg9b56gIVWe3tCh13gg4pEAYYByABEgKThPD_BwE

Regarding current limiting for the astable: add an emitter resistor to Q3 & Q4 and clamp the base voltages, with diodes.

[Telek]

Thank you both Zero999 and Ian.M for your input! This is very useful.

It looks like this is way more complicated than I was hoping. I was hoping to stay simple but I want the most longevity and least ghosting, so what do you recommend? I'd prefer to stay with DC in, but I guess I don't have to go transformerless. Ideally 5VDC in as the HV518/HV5812 chips that I'm already using are 5V logic.

Is Rolo's universal VFD PSU with the LM4871 from 7-5-2018 the best bet then? I see there was still some discussion even after that.

Having said that, the SN6501 looks very simple - would that be the best way to go? If so, any hints about the transformer? This is well outside of my wheelhouse. 75mA and 5.6Vac RMS 50-60Hz output is the recommended for the VFD based on the datasheet.

[Zero999]

Thank you both Zero999 and Ian.M for your input! This is very useful.

It looks like this is way more complicated than I was hoping. I was hoping to stay simple but I want the most longevity and least ghosting, so what do you recommend? I'd prefer to stay with DC in, but I guess I don't have to go transformerless. Ideally 5VDC in as the HV518/HV5812 chips that I'm already using are 5V logic.

Is Rolo's universal VFD PSU with the LM4871 from 7-5-2018 the best bet then? I see there was still some discussion even after that.

Having said that, the SN6501 looks very simple - would that be the best way to go? If so, any hints about the transformer? This is well outside of my wheelhouse. 75mA and 5.6Vac RMS 50-60Hz output is the recommended for the VFD based on the datasheet.

The SN6501 is a good idea and works at 380kHz. Don't worry about the frequency. It's just a heater and will work from any frequency.

If you want something very simple, here's a cut down version of my circuit. Although not good practise, it will work without base-emitter resistors, so they can go and the resistor values increased to reduce the surges. It will still need a good decoupling capacitor though.

[Ian.M]

I'm concerned you are designing yourself into a corner here. 

From Noritake's guide: "...  reduction of filament voltage or anode/grid voltage is not recommended because this may cause uneven illumination."  Your filament specs: 5.6V RMS @ 75mA.

However you are looking at BJT H-bridges, which are going to drop 0.5V or even more (to get 2x Vce_sat under 0.5V would require much larger transistors than you are considering and a lot more base drive), so starting from a nominal 5.0V Vcc rail, at best you'll get 4.5V RMS to the filament, underrunning it by 20%.   Cranking up the HT in an attempt to compensate for low emission is *NOT* recommended, as that accelerates ageing of both the phosphor and the filament coating by ion bombardment.

Even with a MOSFET H-bridge powered from your nominal 5V Vcc, the filament voltage will be marginal unless you run your Vcc rail a bit on the high side.

If you only have 5V in from a 'USB' power supply, you've *really* boxed yourself into a corner.  The only ways out are a boost converter to power the filament driver, or to wind a custom step-up filament drive transformer for a Royer oscillator filament driver.  However custom magnetics are expensive and difficult to prototype even with a suitable pot-core development kit and considerable coil-winding experience.

[Zero999]

I'm concerned you are designing yourself into a corner here. 

From Noritake's guide: "...  reduction of filament voltage or anode/grid voltage is not recommended because this may cause uneven illumination."  Your filament specs: 5.6V RMS @ 75mA.

However you are looking at BJT H-bridges, which are going to drop 0.5V or even more (to get 2x Vce_sat under 0.5V would require much larger transistors than you are considering and a lot more base drive), so starting from a nominal 5.0V Vcc rail, at best you'll get 4.5V RMS to the filament, underrunning it by 20%.   Cranking up the HT in an attempt to compensate for low emission is *NOT* recommended, as that accelerates ageing of both the phosphor and the filament coating by ion bombardment.

Even with a MOSFET H-bridge powered from your nominal 5V Vcc, the filament voltage will be marginal unless you run your Vcc rail a bit on the high side.

If you only have 5V in from a 'USB' power supply, you've *really* boxed yourself into a corner.  The only ways out are a boost converter to power the filament driver, or to wind a custom step-up filament drive transformer for a Royer oscillator filament driver.  However custom magnetics are expensive and difficult to prototype even with a suitable pot-core development kit and considerable coil-winding experience.

Well he said:

Some preliminary tests seem to indicate that this works perfect! I'll add some pics in another post. I don't have 1.5kOhm resistors, so I used 1k+440 and a 1uF 400V cap (it's what I had), all other components as you described. I am getting about 600Hz under load with 4.68VDCin and 4.40VAC (cyc RMS) out and 9.72V pk-pk. Filament draw is 71mA AC. According to the datasheet, typical at 5.6Vac 60Hz RMS is 73mA, so that tells me that I have it dialled in nice.

So there must be a fair tolerance on the figure of 5.6VAC, which is given as a typical figure and 71mA is only slightly lower than the  data sheet.

If the voltage is too low, a Royer converter or a transformer driver IC are a good ideas, but there's the added expense of a transformer.

[Telek]

I'm concerned you are designing yourself into a corner here. 

From Noritake's guide: "...  reduction of filament voltage or anode/grid voltage is not recommended because this may cause uneven illumination."  Your filament specs: 5.6V RMS @ 75mA.

However you are looking at BJT H-bridges, which are going to drop 0.5V or even more (to get 2x Vce_sat under 0.5V would require much larger transistors than you are considering and a lot more base drive), so starting from a nominal 5.0V Vcc rail, at best you'll get 4.5V RMS to the filament, underrunning it by 20%.   Cranking up the HT in an attempt to compensate for low emission is *NOT* recommended, as that accelerates ageing of both the phosphor and the filament coating by ion bombardment.

Even with a MOSFET H-bridge powered from your nominal 5V Vcc, the filament voltage will be marginal unless you run your Vcc rail a bit on the high side.

If you only have 5V in from a 'USB' power supply, you've *really* boxed yourself into a corner.  The only ways out are a boost converter to power the filament driver, or to wind a custom step-up filament drive transformer for a Royer oscillator filament driver.  However custom magnetics are expensive and difficult to prototype even with a suitable pot-core development kit and considerable coil-winding experience.

You guys definitely know more than me So I'm happy to change the design based on your feedback. I want to do this right, but with a preference towards simple. If we can do it simpler and still right, that's where I want to go, but if we need to keep it less simple then I'm ok with that.

As Zero999 pointed out, and with my understanding, the current should be the critical factor overall. So I think that 5V in with his circuit should be ok, but I'm happy to change if that's the consensus. Also note that I did measure 5.01VDC in and 4.77Vac out with 71mA.

I know that powering off 5V USB is limiting, so if that's not really feasible then we can switch it up. However, another thought - since I have to boost to ~30V anyway for the segments, I wonder if I can still use the 5VDC in, then from the 30VDC segment use a 1:5 step-down transformer from there could work with something like the SN6501 to give me simple and relatively off-the-shelf without needing custom drivers / transformers? The boost converters that I already have are 3A, and I don't need anywhere near that so there is room there.

The datasheet calls for 35Vp-p (max 42) for the grid/anode but again they're expecting AC. If I provide DC, I'm thinking somewhere around ~30VDC is the correct spot. But I think there's some flexibility there, so that way I can stick with a 1:5 transformer and still tweak it to get the correct Vac out.

[Ian.M]

I notice you haven't provided a link to the VFD datasheet, nor have you stated whether or not your MCU is capable of generating a 50% PWM at the desired multiplexing frequency or an integer multiple of it.

What does the filament look like if you power it from an adjustable DC source and slowly crank the voltage up until you either reach its nominal voltage or nominal current (5.6V or 75mA, if I'm not mistaken)?  DO *NOT* continue increasing the voltage if it starts glowing brighter than a dull red. It shouldn't be bright enough to be noticeable in normal diffuse domestic room lighting.  If its glowing noticeably red before you reach its nominal ratings, you may well be able to get away with a lower filament voltage without issues. 

42V is less than 10x your 5V supply, so a simple single inductor boost converter should be adequate for the HT supply - no transformer required. I'll defer further comments on the filament supply till I've seen the VFD datasheet, and you've seen how bright the filament gets as you approach its nominal voltage and current.

This article on the effect of filament voltage on transmitter tube operating life is rather interesting: https://www.w8ji.com/filament_voltage_life.htm

[Telek]

I notice you haven't provided a link to the VFD datasheet, nor have you stated whether or not your MCU is capable of generating a 50% PWM at the desired multiplexing frequency or an integer multiple of it.

I was (quite surprisingly) sent the datasheet directly from Noritake, I have asked if I'm allowed to share it publicly and they are apparently discussing it, but haven't given me a yes or a no yet.  I'm using an Arduino right now, and will be using an ESP8266 later, so yes to the second part of your question. Let me type up the critical bits in a spreadsheet and I'll share it.

What does the filament look like if you power it from an adjustable DC source and slowly crank the voltage up until you either reach its nominal voltage or nominal current (5.6V or 75mA, if I'm not mistaken)?  DO *NOT* continue increasing the voltage if it starts glowing brighter than a dull red. It shouldn't be bright enough to be noticeable in normal diffuse domestic room lighting.  If its glowing noticeably red before you reach its nominal ratings, you may well be able to get away with a lower filament voltage without issues. 

I am currently using an adjustable DC regulator from 12VDC in, and have done exactly that previously. At 5.6VDC even in a fairly dark room I cannot see the filaments at all. Around 6.5VDC I can see a faint red glow. At 7.4VDC I can definitely see a clear red glow (two photos attached).  This was done a few weeks ago, and I have 15 of these so I wasn't too caring if I burnt one out. I didn't check the currents at those limits though.

At 5.11VDC source right now (tuned to 5.01V with no load) and 4.84VDC measured across the filament legs, I'm seeing it settle to 68.7mA after a few seconds. At this level, the leftmost (brightest) segments on the display are quite bright even in a bright room, with a clear drop-off to the right. I'm sharing the 5VDC with logic chips, so I don't want to bump it up too much past this right now, but I don't think that I really need it any brighter.

I've attached a photo at 5.11VDC source with the fade. It's difficult to really capture it. I wrote a small app that goes through all 96 characters in my font, moving one column at a time, but "building up" the columns from the right.

As per your question, I slowly dialed it up. At 5.61VDC source and 5.44V measured across the filament legs, I have 73mA which is exactly what the spec sheet says should be nominal. I don't want to go above that right now unless I separate out the filament voltage.

I haven't done any brightness testing yet below 5V though. Should I?

42V is less than 10x your 5V supply, so a simple single inductor boost converter should be adequate for the HT supply - no transformer required. I'll defer further comments on the filament supply till I've seen the VFD datasheet, and you've seen how bright the filament gets as you approach its nominal voltage and current.

This article on the effect of filament voltage on transmitter tube operating life is rather interesting: https://www.w8ji.com/filament_voltage_life.htm

I am currently using a cheap ebay/Amazon generic step-up regulator https://www.amazon.ca/dp/B07JNQFV7F/ and it seems to be working well at 30VDC from 12VDC in.

Thank you very much for that article, I will read it, and thank you very much as well for helping me out here! I really appreciate it!

[Telek]

Here's the relevant section of the datasheet

[Zero999]

Here's the relevant section of the datasheet

According to that, 4.8V is below the minimum, so is marginal design. How many are you going to build? If just is just a one off, then it's clearly working, so stick with it. On the other hand, if you're mass producing it, then you should use a higher voltage, as it might not work for all displays. You could add a buck converter and h-bridge, after the boost converter, but it's messy.

Another possibility is to use a transformer for the booster, with another winding for the filament supply.

[Telek]

Here's the relevant section of the datasheet

According to that, 4.8V is below the minimum, so is marginal design. How many are you going to build? If just is just a one off, then it's clearly working, so stick with it. On the other hand, if you're mass producing it, then you should use a higher voltage, as it might not work for all displays. You could add a buck converter and h-bridge, after the boost converter, but it's messy.

Another possibility is to use a transformer for the booster, with another winding for the filament supply.

I have about a dozen that I'll put together for various personal use and gifts. Since my goal is to drive with the ESP8266 I want to make an "alarm clock" type of thing with some buttons on it, and have it auto connect to some wifi points. Then I can control/send them messages remotely. Give some to friends as well.

What if I just used a step-down from the 30V anode voltage regulator? That way I don't have to worry about Vin. Not the most efficient, but easy to do.

Note that I have NOT tried this yet with your filament circuit, I'm just straight DC driving it at the moment. I wanted to get my code working first before trying to play with the filament drive (otherwise I wouldn't have anything to compare it to). I'm still tweaking the code.

I should mention, now that I actually have the code a bit better and not using just a flickering test, the brightness is a lot lower than I would expect. I'm trying to tweak the settings to not get flicker, so maybe I'll figure out how to get it brighter. Note that the photo makes it look worse in terms of glow. In reality it's very clean right now, no ghosting or background illumination... just dimmer than I was expecting.

Edit - ok, this is odd. So I used the Vusb for logic to separate out the filament. Running from 3.0 VDC to 6.5 VDC for filament seems to have no impact on brightness. Maybe I need a higher anode voltage? Bringing that up to 35VDC certainly did make it brighter, but still not as bright as I expected based on 100% duty (all segments and columns on). 25VDC was quite a bit dimmer.

[Telek]

Update, the original astable self oscillating h-bridge is working well. This is now well into the most complicated breadboard system I've put together



Measured, 5.03VDC in under load, 4.87VAC Cyc RMS shows on oscilloscope, with 10.2Vp-p at ~505Hz +/- 10. Hovers right around 75mA under full operation (+/- 0.5mA). The waveform stabilized (less jittery) under full operational load.



I have not added inductors yet, and I don't have any 2Ohm resistors to work with (although I probably have some 1/4W 1 ohm ones, I guess I could tie 2 together to test). I do have some 1812 SMD 22uH though, should I try one each of those in series with the positive emitters to help? Or would you still prefer one of the two later designs? I'm happy with this one so far, but I don't know what it means for the longevity of the VFD.

I had an error in my original code due to misunderstanding some awkward wording in the datasheets. Now, at 30VDC grid, and no awkward delays to avoid the bad ghosting I had before, with no blanking, the brightness is quite acceptable and the background is quite clean. I don't know if I need to do any blanking. There's the tiniest amount of ghosting from the previous column, but I have to look hard to find it.

Noritake did finally get back to me, they have asked me not to share the datasheet (the previous snip was what I retyped myself), but they are interested in the results of my tinkering. If anyone is interested they welcome to ask them on Twitter for the datasheet directly and they will provide.

Oddly, with my oscilloscope attached and the computer USB attached, something strange happens. The waveform gets destroyed and almost entirely negative, and an odd buzzing comes from the 5V switching regulator. The brightness also increases ever so slightly. This only happens if the USB cable is connected, or even if only the USB ground is connected to circuit ground  .



If I disconnect the oscilloscope ground, the perfect waveform returns at only half strength (2.5VAC but the display brightness doesn't change). It's the same as if I probe either side of the bridge to ground, instead of across both sides of the bridge.

I also found that one of my DS3231 boards appears to be a cheap knockoff and defective (not that I expected real ones when buying them on Amazon for $2ea) and my Arduino nano internal oscillator seems to be off by enough to be 0.14sec slow every minute... I'm starting to think that maybe I should separate this out into its own thread 

[Zero999]

Which circuit did you build?

Regarding the oscilloscope: it sound like a ground loop issue.

[Telek]

Which circuit did you build?

Regarding the oscilloscope: it sound like a ground loop issue.

This is the one that I built.

Yeah, I figured some sort of ground loop. I tried to add a few decoupling caps but no dice. I have a UPS so just ran it off there for a few minutes to check that it was ok lol.

[Telek]

If I do decide to go the IC route - what do we think of something as simple as TC4428 (high-speed 1.25 dual mosfet, one inverting). Run it off PWM from the MCU and drive the filament directly from the outputs. If I'm reading all this correctly, it's like a simpler BD6211 without the protection diodes needed for a motor, and the extra control pins. I could just PWM to both IN1 and IN2 from the same pin, since it's one inverting.

With the astable route, is it possible to tweak one side of it to get it so that the negative side is half the voltage and twice the time? As per the Noritake site:



I assume change R4 on the negative side of the capacitor?

[Zero999]

If I do decide to go the IC route - what do we think of something as simple as TC4428 (high-speed 1.25 dual mosfet, one inverting). Run it off PWM from the MCU and drive the filament directly from the outputs. If I'm reading all this correctly, it's like a simpler BD6211 without the protection diodes needed for a motor, and the extra control pins. I could just PWM to both IN1 and IN2 from the same pin, since it's one inverting.

No, because the TC4428 has far too higher output resistance, of around 7 Ohm on each output, so the voltage drop will be too high.

With the astable route, is it possible to tweak one side of it to get it so that the negative side is half the voltage and twice the time? As per the Noritake site:

(Attachment Link)

I assume change R4 on the negative side of the capacitor?

No, getting asymetrical output voltages would need a much more complicated circuit. I don't know why you'd want to drive the filament, with that waveform.

[Ian.M]

The text next to that figure states "However, a 1/2 duty factor should be set  ..."

The *ONLY* reason to use an asymmetrical pulse drive waveform for the filament, is that when AC coupled, it lets you reduce the RMS voltage below that a 50% duty cycle waveform would give, without loosing efficiency.   

LTspice sim of 5% to 95% duty cycle AC coupled PWM with RMS measurement attached.  Right click the error log that pops up and Plot the .step'ed .meas data, for Vrms vs duty cycle.

The disadvantage of asymmetric drive waveforms and waveforms with higher crest factor, is you need a higher HT for the same brightness as you need to increase the filament bias voltage to prevent ghosting on positive going filament supply peaks.

[Telek]

If I do decide to go the IC route - what do we think of something as simple as TC4428 (high-speed 1.25 dual mosfet, one inverting). Run it off PWM from the MCU and drive the filament directly from the outputs. If I'm reading all this correctly, it's like a simpler BD6211 without the protection diodes needed for a motor, and the extra control pins. I could just PWM to both IN1 and IN2 from the same pin, since it's one inverting.

No, because the TC4428 has far too higher output resistance, of around 7 Ohm on each output, so the voltage drop will be too high.

With the astable route, is it possible to tweak one side of it to get it so that the negative side is half the voltage and twice the time? As per the Noritake site:

I assume change R4 on the negative side of the capacitor?

No, getting asymetrical output voltages would need a much more complicated circuit. I don't know why you'd want to drive the filament, with that waveform.

  just going based off the manufacturer's website https://www.noritake-elec.com/technology/general-technical-information/vfd-operation

Ok, so based on how this circuit is working well, do I just add the inductors in-line with the emitters of Q3/Q4?



EDIT: Hmm. Here's what the waveform looks like with no inductors:



and with two inductors (I found some nice 0.57uH 1.5ohm through-hole inductors in my parts bin) from the +5V rail to emitter.



The voltage drop I expected due to the resistance, which pulls it a bit lower than I'd like. But is that expected for the waveform? Is that waveform a problem? I noticed that I can get inductors with much less resistance - should I try for maybe 33uH 0.1ohm?

Interesting thing to note - all this runs perfectly fine off USB power right now, with expected brightness. Is this maybe something that if it's working, don't worry about it?

I don't know why this editor just doesn't let me get those attachments working inline.

[Telek]

The text next to that figure states "However, a 1/2 duty factor should be set  ..."

The *ONLY* reason to use an asymmetrical pulse drive waveform for the filament, is that when AC coupled, it lets you reduce the RMS voltage below that a 50% duty cycle waveform would give, without loosing efficiency.   

LTspice sim of 5% to 95% duty cycle AC coupled PWM with RMS measurement attached.  Right click the error log that pops up and Plot the .step'ed .meas data, for Vrms vs duty cycle.

The disadvantage of asymmetric drive waveforms and waveforms with higher crest factor, is you need a higher HT for the same brightness as you need to increase the filament bias voltage to prevent ghosting on positive going filament supply peaks.

Neat, thanks! I figured it had something to do with not burning out the filament or something. If a 50% duty cycle works, let's not complicate it!

[Zero999]

If I do decide to go the IC route - what do we think of something as simple as TC4428 (high-speed 1.25 dual mosfet, one inverting). Run it off PWM from the MCU and drive the filament directly from the outputs. If I'm reading all this correctly, it's like a simpler BD6211 without the protection diodes needed for a motor, and the extra control pins. I could just PWM to both IN1 and IN2 from the same pin, since it's one inverting.

No, because the TC4428 has far too higher output resistance, of around 7 Ohm on each output, so the voltage drop will be too high.

With the astable route, is it possible to tweak one side of it to get it so that the negative side is half the voltage and twice the time? As per the Noritake site:

I assume change R4 on the negative side of the capacitor?

No, getting asymetrical output voltages would need a much more complicated circuit. I don't know why you'd want to drive the filament, with that waveform.

  just going based off the manufacturer's website https://www.noritake-elec.com/technology/general-technical-information/vfd-operation

Ok, so based on how this circuit is working well, do I just add the inductors in-line with the emitters of Q3/Q4?



EDIT: Hmm. Here's what the waveform looks like with no inductors:

[ Specified attachment is not available ]

and with two inductors (I found some nice 0.57uH 1.5ohm through-hole inductors in my parts bin) from the +5V rail to emitter.

[ Specified attachment is not available ]

The voltage drop I expected due to the resistance, which pulls it a bit lower than I'd like. But is that expected for the waveform? Is that waveform a problem? I noticed that I can get inductors with much less resistance - should I try for maybe 33uH 0.1ohm?

Interesting thing to note - all this runs perfectly fine off USB power right now, with expected brightness. Is this maybe something that if it's working, don't worry about it?

I don't know why this editor just doesn't let me get those attachments working inline.

There have been a few problems with attachments on the forum recently, so Dave has disabled inline attachments. It doesn't matter, as I can see all of the images you've attached, as expandable thumbnails. If you still want inline images, click on the image to expand it, copy the URL to the clipboard, go back to the post and past it into [img][/img] tags.

If it works, then leave it be. Don't worry about the waveform.

The inductor was just to limit the current spike. Only one might be needed, but unfortunately you won't know whether it's required at this stage or not, until you build the entire circuit. My advice is to leave room for it on the PCB, so it can be added, if the MCU is unstable, or replaced with a wire link, if it works. You could also design the PCB for the  current limiting design, with extra pads for the inductor.

[Telek]

If it works, then leave it be. Don't worry about the waveform.

Great! It occurs to me now that the waveform is exactly what we'd expect with two inductors, one per transistor. If I use only one shared inductor then the waveform should be normal as current is constantly flowing.

The inductor was just to limit the current spike. Only one might be needed, but unfortunately you won't know whether it's required at this stage or not, until you build the entire circuit. My advice is to leave room for it on the PCB, so it can be added, if the MCU is unstable, or replaced with a wire link, if it works.

That's a great idea and exactly what I'll do. Does it make sense to look for an inductor with a very low DC resistance so that I don't get the voltage drop? My understanding is that the current limiting capabilities are based on the EMF buildup, so the pure DC resistance shouldn't impact that? So like 33uH and 0.15ohm? Or 10uH and 0.17ohm? Or 47uH 0.17ohm? Since it's pretty good on its own, I assume any of these will be fine. Fun fact: USB from my laptop works fine, but from a 7-port external powered USB hub it has an overcurrent situation. I assume the spike when it plugs in is just a bit too much for the hub to handle. I'm thinking this one: https://www.mouser.ca/ProductDetail/Bourns/RLB0712-330KL

You could also design the PCB for the  current limiting design, with extra pads for the inductor.

What's the purpose of the diodes in that one? I know you mentioned that it clamps the base voltages, but I'll have to admit that I don't know what that means.

[Zero999]

Yes, a low ESR inductor is a good idea.

Regarding the diodes:
The base voltage of a PNP transistor needs to be about 0.6V below the emitter voltage, in order for it to turn on. As the current rises, the voltage across R5 also increases, pushing the emitter voltage on Q3 & Q4 down, with respect to the supply voltage. When the current gets too high, the diodes clamp the base voltages, cutting off the current flow.

I hope my explanation is good enough. If not, I'll post more tomorrow. It's getting late.

[Nifty]

I've had another idea. Since the frequency isn't important. It's possible to use an astable multivibrator to drive it. Two additional transistors can be added to the typical astable circuit to form a self oscillating h-bridge.
http://en.wikipedia.org/wiki/Multivibrator#Operation

Hi! Dear Sir, In the schematics of the cassette deck that I am currently restoring (TEAC V-770), I saw a very similar solution.
Unfortunately, I am not well in analog circuitry.
Can you please help me to understand, how can I slightly increase the voltage at the output of this generator.


https://ibb.co/x1rXDND

Thank you!