Design of SMPS - Not sure what these spikes are being caused by.

[Zog]

Ok .. I now have 125 mA's out as per your instructions Tim,
Well done on picking that up for me.

The gate still shows ringing of course but the FET run's very cool now which is a big plus.
Zero missing pulses.

The guys design runs at slightly below the recommended chip frequency.
I might try to up that a bit. His reasoning was the gate charge even though pretty low would lower the efficiency.
What's your opinion ? I would rather up it and use smaller inductors.

I am now experimenting with different winding ratios and will report back.
Once I have got that down I will start implementing the rest of your suggestions.

After that I will continue with the suggestions on layout posted before.
The dirty ground net naming trick is a good one I have never seen anywhere else.

You were also right about that other signal. It is all over the board anywhere you poke the probe. Layout of course.

Cheers mate.

[T3sl4co1l]

Gate charge is a tiny part of overall switching loss, far more significant is the stray inductance causing RF to blast out of everywhere!

As the leakage inductance is part of that switching loss (unless clamped and recycled -- an advanced topic), you can't reduce losses terrifically far in this type of design.  A good transformer is required.  Still, it should be able to run at 200kHz easily, given improved layout and an okay transformer.

Adding gate resistor(s) is necessary with the layout as-is, as this reduces the switching speed.  You must slow it to below the rate of that ringing, so as not to excite it.  Then your waveforms will look alright.

Tim

[Zog]

Thanks Tim,
Yes a gate resistor was mentioned by someone else too.

Is that in addition to the snubber on the gate or can that be the "gate" resistor. I am assuming it can be/is.

BTW A Zero poster called Carlotta sent me a PM pushing their "prototyping" service.
 https://www.smart-prototyping.com/. You can also contact me at carlota@smart-prototyping.com
Anyone else heard of this crowd ?.
I am pretty happy with my current manufacturer, so doubt I will change.

[T3sl4co1l]

Thanks Tim,
Yes a gate resistor was mentioned by someone else too.

Is that in addition to the snubber on the gate or can that be the "gate" resistor. I am assuming it can be/is.

Some kind of gate resistance, yes.  If it's two resistors and a diode, or just a resistor, whatever, that's just different ways of getting there, more options.

BTW A Zero poster called Carlotta sent me a PM pushing their "prototyping" service.
 <snip free advertising> You can also contact me at carlota@SPAM
Anyone else heard of this crowd ?.
I am pretty happy with my current manufacturer, so doubt I will change.

Same.  Mods, if you would please -- discipline said account for PM spam?

Tim

[Zog]

Roger roger,
Thanks Tim.

I will "go dark" on this topic for a few weeks now.
Need to get a new board layout done for testing.
Source parts etc.

Will be back once this is done.
Cheers,
Phill

[Zog]

Hi All,
I have given up trying to make this myself and need a pro to do it for me, preferably in Eagle. (paid job)
Need 100 Watts.
Anyone interested ?

[ocset]

Your vout is 185VDC
Your Pout is 100w
Your vin is =12v?

Have you thought of the dual cascaded boost?
Ill  send you some stuff on it if you want.

(I am working through the three pages to see where you got stuck....i saw you were wondering about winding a tapped inductor at one point, and at one point you were speaking of a noise issue.
, you gave a website of a cct you "copied"?...
https://jan.rychter.com/high-voltage-power-supply-for-nixie-tube-projects
)

[MagicSmoker]

Hi All,
I have given up trying to make this myself and need a pro to do it for me, preferably in Eagle. (paid job)
Need 100 Watts.
Anyone interested ?

Is this still for Nixie tubes, which require approximately 10mA each? So 100W would be good for around 50 of them?!?

Input voltage? Size constraints? Budget? I likely have something in my pile of past projects that can be easily modified for the job, and I use EAGLE 7.7 (ie - the last version before getting Borg'ed by Autodesk), but I am a bit busy for the next couple of weeks.

[ocset]

If you give the input voltage, then ill send you a design, and you can see if you want to go on and do it yourself, or contact myself.

PCB layout doc attached...did you follow these rules?

I must admit im an eagle 7.4.0 user.

Also,  there is the cascaded boost, which i could modify for you if you wish...its as attached

[Zog]

@treez

If you give the input voltage, then ill send you a design, and you can see if you want to go on and do it yourself, or contact myself.

I don't have the skills to try to do it myself.

@treez

Code: [Select]

Also,  there is the cascaded boost, which i could modify for you if you wish...its as attachedOnly text in that file ?

@MagicSmoker

Code: [Select]

Input voltage? Size constraints? Budget? I likely have something in my pile of past projects that can be easily modified for the job, and I use EAGLE 7.7 (ie - the last version before getting Borg'ed by Autodesk), but I am a bit busy for the next couple of weeks
Input voltage can be from 5VDC to 24VDC. Size ? about the size of a packet of cigarettes. Output Voltage 220VDC because I want to drive a doubler circuit to get 440VDC for Dekatrons.

[MagicSmoker]

...
Input voltage can be from 5VDC to 24VDC. Size ? about the size of a packet of cigarettes. Output Voltage 220VDC because I want to drive a doubler circuit to get 440VDC for Dekatrons.

A nearly 5:1 input voltage range is absurdly impractical. The only commercial supplies that accommodate anywhere close to that are universal input flybacks that can run on 85VAC to 265VAC, and even that is "only" a 3:1 range. Cramming 100W into the volume of a cigarette box is also going to be challenging, especially with the input voltage restriction.

Do you need the 220V and 440V at the same time, or is it an either/or thing?


EDIT - fat-fingered posting before finishing.

[ocset]

yes, 100W from a 5V input means >20A of input current. Size may need to be bigger, but if it can be a wide pack of cigarettes, then we may be able to get near it.

Only text in that file ?

(The .txt file you can change to .asc, then  open in the free LTspice simulator)

[Zog]

A nearly 5:1 input voltage range is absurdly impractical. The only commercial supplies that accommodate anywhere close to that are universal input flybacks that can run on 85VAC to 265VAC, and even that is "only" a 3:1 range. Cramming 100W into the volume of a cigarette box is also going to be challenging, especially with the input voltage restriction.

Sorry I didn't make myself clear.
That voltage range is what I can deal with in terms of getting another small  DC power supply to drive it.
In other words I would be getting a separate commercial one to convert 110/240VAC to the DC provided for this design. I was thinking of just buying a meenwell or other reputable brand for the DC input.

Do you need the 220V and 440V at the same time, or is it an either/or thing?

I need both at the same time.

[Zog]

I just found this one on ebay. I have not done a search for a while.
eBay auction: #https://www.ebay.com.au/itm/High-Voltage-DC-DC-150V-420V-Converter-NIXIE-Tube-HV-Power-Supply-1-25-12V-adj/333261626043?hash=item4d97f0dabb:g:UKMAAOSwPbldTjVL
Could I just put two or more of these in parallel to get the mA's I need ? Is that possible ? Common mode problems ?
Might be a quick and dirty way to do it ?

[Zog]

yes, 100W from a 5V input means >20A of input current. Size may need to be bigger, but if it can be a wide pack of cigarettes, then we may be able to get near it.

I don't have a preference for input voltage.  Looking at a mean well supply 12V or 24V DC to power it. They are quite small and are high wattage.

Only text in that file ?
(The .txt file you can change to .asc, then  open in the free LTspice simulator)

LT Spice is wayyy beyond me !

[jmelson]

The spikes in the middle of the ramp may be real.  There are some other, taller, spikes that are very narrow.  These may not be real.
First check is to connect the probe ground to your normal ground point, and then clip the probe tip to the same point.  if you see a bunch of spikes, these are coming through the GROUND of the probe, NOT through the signal pin of the probe.  That causes current through the ground shield of the probe.  You need to improve the ground between the ground point on the scope and the ground point of the equipment under test.

Jon

[Zog]

The spikes in the middle of the ramp may be real.  There are some other, taller, spikes that are very narrow.  These may not be real.
First check is to connect the probe ground to your normal ground point, and then clip the probe tip to the same point.  if you see a bunch of spikes, these are coming through the GROUND of the probe, NOT through the signal pin of the probe.  That causes current through the ground shield of the probe.  You need to improve the ground between the ground point on the scope and the ground point of the equipment under test.

Thanks for that. It was pointed out before but I appreciate your input.

[Zog]

This image is a little unclear.
Anyone have any idea what the main driver chip is ?

[T3sl4co1l]

TL494 or similar.

Tim

[Zog]

Tim,
This is the circuit I came up with after you supplied me those hand drawn sheets.
Does this look right to you in general connections terms  ?

[T3sl4co1l]

C8 charges when Q2 turns on, which will probably defeat the peak current mode operation of the controller.

They also don't document it very well, but TPS40210 goes into hiccup mode (/SS pulls low, stays off until crosses rising threshold) rather than current limiting.  It's rather dumb.

The combination will probably frustrate operation of this circuit enough that it never starts up, or doesn't maintain the full output you were expecting, or isn't able to recover from a momentary overload condition.

You can use multiple secondaries as their own flyback outputs, and connect the DC outputs in series.  Then you don't need C8 charging on the "down stroke", energy gets delivered to the output only in the flyback period (Q2 off).  Probably use N = 2-3 of say MSS12xx dual-winding inductors, with the primaries in parallel, and the secondaries wired accordingly.  (Choose inductance to be N times higher per inductor, because for uncoupled inductors, the value reduces in parallel.)

Tim

[Zog]

Thanks Tim,
That explains why it never started up.
Will look at implementing your new suggestions.

The circuit is wired correctly though right ?

You can use multiple secondaries as their own flyback outputs, and connect the DC outputs in series.  Then you don't need C8 charging on the "down stroke", energy gets delivered to the output only in the flyback period (Q2 off).  Probably use N = 2-3 of say MSS12xx dual-winding inductors, with the primaries in parallel, and the secondaries wired accordingly.  (Choose inductance to be N times higher per inductor, because for uncoupled inductors, the value reduces in parallel.)

Could you perhaps provide a hand drawn circuit explaining to us non engineers just what you mean by that ?

C8 charges when Q2 turns on, which will probably defeat the peak current mode operation of the controller.

It did not do that on the original circuit ? Little confused now ?    (original is the original post)

[T3sl4co1l]

Other than that, yeah I think so.

There was an appnote I think from TI's "flybuck" documentation that showed the connection, I can't seem to find it offhand.  Try giving that a look.

Yes, the original circuit would have the same problem.

Tim

[Zog]

I thought these files might help other beginners like me at how to layout SMPS. treez layout instructions are very good. https://www.eevblog.com/forum/projects/design-of-smps-not-sure-what-these-spikes-are-being-caused-by/msg2622906/#msg2622906

The TPS40210 Datasheet http://www.ti.com/lit/ds/slus772f/slus772f.pdf is good but kind of difficult to follow given that the example circuit does not match the layout. The example layout is just the bottom and the top, so I got creative in photoshop and merged them together and added component references so it's a little easier to follow.
Pretty sure it's a hexfet package used for Q1.





Top and Bottom layers merged.


Top Layer


Bottom Layer


I do have a question though.
What are the R1,R2 and the jumper for ?

[T3sl4co1l]

Looks to be routed to EN. Don't know why it's not on the schematic.

Tim