Oscope - Roman Black's Two Transistor Switcher Design

[ericgibbs]

Eric:
I thought the experiment was going to be to compare the two circuit waveforms side by side, ie with no battery (connected normally) and then with the battery instead of the emitter connected to the output?
Can you show the waveform with and without the battery? I think that would do it. They should be similar.

Hi Al,
Sorry I misunderstood, will do the above and post later.

E.

EDIT:
Upper plot group is the battery version, lower group the original circuit.

If you need more detail, just ask.

 

[MrAl]

Very cool effect! Thanks MrAl and EricGibbs for testing that, it's interesting to see it will still oscillate based on collector current (collector saturation) of Q1 plus possibly some inductor saturation effect in there too.

However I think this will require higher peak inductor current (with the associated loss in efficiency) and of course the very real issue that there is no load regulation!

In one of my early tests before this design I had an extra resistor of 1 ohm or so between the inductor output and the load cap. Feedback was taken from the front of the resistor to Q2 base, so there was a much larger voltage ripple to activate Q2.

Once the RC delay was added there was no longer a need for the load voltage to have ripple to give a reliable low freq oscillation.

Hi Roman and Eric,


Roman:
Well actually it should work the same way as with the small battery except for the load or other regulation. The idea is not to show how it regulates, just how it turns on and off. The regulation should be a slower process so that the regulation cant increase the ripple voltage on the output.

If you look at Eric's new plot you'll see a variation in the inductor current which should not be there. We'll have to see if Eric did a shot too soon in the start up sequence first though. We can get rid of that if it continues even after a decent start up time.

Eric:
Can you repeat that but take the time shot after say 10ms ? We're trying to see the operation after a decent start up period after all the active elements have reached their nominal values.
If that instability persists, we can get rid of it quite easily i think. The instability creates problems one being increased output ripple...but that's only if it persists after a decent start up time period.

 

[tvtech]

This is way beyond me guys..

At least I have the balls to admit it

Love the lot of you

tvtech

 

[ericgibbs]

Eric:
Can you repeat that but take the time shot after say 10ms ? We're trying to see the operation after a decent start up period after all the active elements have reached their nominal values.
If that instability persists, we can get rid of it quite easily i think. The instability creates problems one being increased output ripple...but that's only if it persists after a decent start up time period.

Hi Al,
Those plots are at the stable point, delayed plots by 19mSec.

That jitter on the timing has shown up on most of the sims.

Eric

 

[MrAl]

Hi again Eric,

Can you try adding a 1 ohm resistor in series with the emitter of Q2 ?

 

[ericgibbs]

Hi again Eric,

Can you try adding a 1 ohm resistor in series with the emitter of Q2 ?

hi Al,
Here it is.

 

[MrAl]

Hi Eric,

Ok. There's got to be a reason it is so unstable. Perhaps it is because we are not using any series resistance for the coil?
Maybe 1 ohm in series with that too.

 

[ericgibbs]

Hi Eric,

Ok. There's got to be a reason it is so unstable. Perhaps it is because we are not using any series resistance for the coil?
Maybe 1 ohm in series with that too.

Hi Al,

The inductor has the resistance and inductive values as shown in RB's web page circuit.

Eric

EDIT:
Al,
I believe the jitter is an LTSpice artefact.
The characteristics of the 47uF on the output has an effect on the jitter.
E.

 

[Mr RB]

That instability does not occur in the actual circuit, apart from the situation where output current is too low to really get the inductor and output cap "working". If you check my 'scope photos you can see how stable it is once current is greater than about 15mA.

Also there should be some mV of voltage ripple on the output, I'm curious what you are getting there?

Ideally to tune the circuit the RC delay keeps the SMPS off long enough to get some reduction in Vout, I can't remember exact ripple voltage but 20mV might be ok. Then when the RC delays ends and the regulator is re-enabled the ON time will be longer, as it takes that bit longer to recover the lost Vout and oscillate nicely around the regulated Vout.

I think some of the operation you are seeing is due to reduced Q1 and Q2 gain, and it's oscillating a bit more based on Q1 characteristics and not as much on Vout characteristics as it should, if you get my meaning.

One thing you can try is to set the transistors to BC337-40 and BC327-40, these are the ones I stock and they are the higher beta versions (I think the -40 means beta of 400, so they are a better grade of BC series transistor).

To improve efficiency on most of these designs I was reducing the Q1 base current by increasing R1 and R2, in line with what worked well with the -40 transistors. You will probably get better stability and better Q1 saturation if you increase Q1 base current a bit, especially if you are not using good transistors. Many of my BC transistors measure >500 beta on my beta tester.

 

[MrAl]

Hi Eric and Roman,


Oh ok that's good to hear. Maybe i should fool around with this a bit too. I dont think i have the models for the BC transistors though, i'll have to check. I do have the more standard types like 2N4401 etc.

Roman where are the scope pics you are talking about?

8 or 9 ohms sounds a little high though. Is that the series resistance of the real life circuit too (inductor)?

 

[tvtech]

That instability does not occur in the actual circuit, apart from the situation where output current is too low to really get the inductor and output cap "working". If you check my 'scope photos you can see how stable it is once current is greater than about 15mA.

Also there should be some mV of voltage ripple on the output, I'm curious what you are getting there?

Ideally to tune the circuit the RC delay keeps the SMPS off long enough to get some reduction in Vout, I can't remember exact ripple voltage but 20mV might be ok. Then when the RC delays ends and the regulator is re-enabled the ON time will be longer, as it takes that bit longer to recover the lost Vout and oscillate nicely around the regulated Vout.

I think some of the operation you are seeing is due to reduced Q1 and Q2 gain, and it's oscillating a bit more based on Q1 characteristics and not as much on Vout characteristics as it should, if you get my meaning.

One thing you can try is to set the transistors to BC337-40 and BC327-40, these are the ones I stock and they are the higher beta versions (I think the -40 means beta of 400, so they are a better grade of BC series transistor).

To improve efficiency on most of these designs I was reducing the Q1 base current by increasing R1 and R2, in line with what worked well with the -40 transistors. You will probably get better stability and better Q1 saturation if you increase Q1 base current a bit, especially if you are not using good transistors. Many of my BC transistors measure >500 beta on my beta tester.

Hi guys

Even though I said this discussion was way above me....if circuits rely on specific components meeting A1 rating...as in BC 337-40 (best quality/highest gain)....then the circuit is flawed.

I am familiar with the BC337-40. It was used as the "Impulse Adder" in Telefunken 711 sets that arrived here in 1975 or so.

The sets would ONLY work properly with a genuine BC337-40. 40 being the gain class rating on this little important midget. Put anything else in there with a lower gain..and the set will not work.

Touchy design....but reliable TV's. nevertheless.

When I see "fine tune"..and "electronics" in the same sentence...I worry.

Good designs rely on off the shelf standard components.

True???

Regards,
tvtech

 

[ericgibbs]

hi,
I have tried various combinations of transistors, it works OK in simulation, only a small change in running frequency.

Even tried a 2N3055 and its PNP equivalent, works OK.

As I said in Post #28, with the 47uF having a low internal resistance the frequency is steady.

E.

 

[MrAl]

Hi again,

I tried 2N4401 and 2N4403.
I found that with the 47uf cap having an ESR of 0.01 the circuit was unstable, but increasing that to 0.1 made it stable.
That's with the 50 ohm load.
I also found that taking the 1nf cap out of the circuit it still worked I believe that is because there is a delay introduced with the inductor so the negative feedback is not felt by Q2 until after some small delay period anyway. Interesting. Also interesting is that all instability goes away regardless of the 47uf cap ESR.

I also included the parasitics for the 1nf, and 4.7nf, with an ESR of 0.1 ohm each, and also the input voltage source with 0.1 ohms.

 

[Mr RB]

...
Roman where are the scope pics you are talking about?

There are two scope photos on my SMPS page here;
https://www.romanblack.com/smps/smps.htm
they were pretty typical for the 4 circuits I charted.

...
8 or 9 ohms sounds a little high though. Is that the series resistance of the real life circuit too (inductor)?

Yep that was high, it was the first of the 4 circuits I did, and the inductor was one of those really cheap tiny blue inductors. The other circuits had better inductors. With a good toroid that has low ohms the SMPS will easily top 95% efficiency.

...
Even though I said this discussion was way above me....if circuits rely on specific components meeting A1 rating...as in BC 337-40 (best quality/highest gain)....then the circuit is flawed.
...

That's a good point of course, but I may have given the wrong impression. The circuit will run very well with just about any transistor, I was just saying if Eric was simulating my resistor values etc and trying to match to my results, they were chosen for the specific transistors.

When I mentioned "tuning" the circuit that was in regards to tweaking the efficiency, as far as getting the SMPS working (oscillating reliably etc) then the transistor values are not critical.

...
I have tried various combinations of transistors, it works OK in simulation, only a small change in running frequency.

Even tried a 2N3055 and its PNP equivalent, works OK.
As I said in Post #28, with the 47uF having a low internal resistance the frequency is steady.

Thansk for testing that! I think the instability might be from spice being fussy too, having never seen it with real inductors and caps.

...
I tried 2N4401 and 2N4403.
I found that with the 47uf cap having an ESR of 0.01 the circuit was unstable, but increasing that to 0.1 made it stable.
That's with the 50 ohm load.
...

Interesting! Thanks for that info, I only tried it with electro cap on the output for the web page. I do have a memory of trying a tantalum on the output but can't remember any issues with that.

...
I also found that taking the 1nf cap out of the circuit it still worked I believe that is because there is a delay introduced with the inductor so the negative feedback is not felt by Q2 until after some small delay period anyway. Interesting. Also interesting is that all instability goes away regardless of the 47uf cap ESR.
...

Yep it will still oscillate reliably for most output currents even with no RC delay, it oscillates around the inductor and load cap time constants. But you will find frequency much higher and switching slower, so efficiency goes down quite a bit.

If you (or Eric or Jony130) want a challenge, I was never able to get it working well with very low output currents. For instance 5v 5mA out and 14v 2.5mA in. Which is a bit of a shame as many micro circuits will run from 5mA or so, and having a SMPS which is efficient into 5mA output would be very nice. Maybe for running a 5v or 3.3v micro from a 9v battery?

This is years ago now, but I think it was to do with the inductor requiring a minimum current to be efficient, and/or the transistors. From memory I tried the obvious thing of increasing the resistor values to reduce the current waste compared to the buck current but it just didn't scale well.

 

[()blivion]

Hello, I read all of the other thread on this as well as this one and I am starting to become fascinated about this discussion.

So... I'm simulating *THIS* version of your circuit and I am having some problems. It's lower voltage than it's supposed to be for some reason /).

*THIS* is the simulation, if be so kind to look at it and tell me what exactly is wrong? It's within the current load specs of the schematic on your site, but fails to maintain regulation for some reason. And not just when under extreme load conditions.

Do note that I'm using the "falstad simulator", which is not accurate by any stretch of the imagination, as well as hating some browsers (hint: DON'T use IE). It does work good for simple circuits that you want to analyze in real time. But I'm thinking that although this circuit is "low parts count" it's anything but "simple". It's apparent that it's operation is complex enough to make at least a few people scratch their heads.

Anyway, I could easily see this circuit's intricacies molest the falstad simulator... is all I'm saying.

 

[ericgibbs]

hi,
With that version of the circuit I get this simulation.

I suspect the falstad sim is very limited.

 

[()blivion]

hi,
With that version of the circuit I get this simulation.

I suspect the falstad sim is very limited.

Oh it is.... LOL. Matter of fact, I was just coming back here to say I figured it out. And that it *IS* is the simulator after all. Of course we already knew the circuit worked.

The Fix
To fix the falstad simulator for the above circuit, one simply needs to go into "Options>>Other options..." and turn down the time step to ~10ns, then readjust the simulation speed to compensate for the slow down. I think what was happening is that the NPN (Q2) is doing it's thing so fast that it "falls through the cracks" of the time steps mucking the simulation up. Even some of Paul's built in circuits are made with frequency far far lower than would be the norm. So I expect it's a know limitation. And it would probably also simulate fine if the parts values were butchered enough to drop the frequency without effecting the operation, like some of his circuits. My way is better In any case, with the new settings it gets up to the full 5v like it should, and is even regulated properly. And to my eyes, it works like a text book buck converter. All be it an odd and parts efficient way of going about it.

The Simulator
This wouldn't be the first time that the falstad simulator has let me down. Don't even get me started on it's MOSFET simulation. As I said, I like it for simple circuits where you just want to put it up and make quick adjustments to it. But it fails when accuracy of the simulation is the important thing.

Thanks for simulating it in the real deal. I should stop being set in my ways and learn LTspice. But.... I'm pretty freaking lazy some times.
-()blivion

 

[MrAl]

Hi again,

Oblivion:
LT Spice isnt too hard to learn you'll pick it up fast i bet. There's a lot of people here who could help you with it too if you need it just ask.

Roman:
Yes i assumed that the small cap was there to speed up the switching and thus increase efficiency. I'll have to try to do some tests to see how much it picks up with the caps (and how much it loses without the caps). Should be interesting. With a 9 ohm series resistance in the inductor however i wouldnt count on picking up too much efficiency That probably steals 20 percent right off the top.
I am also assuming a nominal current of 100ma output at 5v or close to that.

 

[MrAl]

Hi again,


Slight error on my part: The circuit doesnt seem to oscillate without the cap C2. What happened was i used a 100k resistor in series with it at first for the test, but then even 100k allows some feedback so it still oscillates. With C2 completely removed from the circuit it is not able to start. This might vary with the actual real circuit because of stray capacitance (even a small amount should be able to start it), but it's not a good thing

Anyway, efficiency is coming in at 87 percent or slightly above like 87.5 percent, and since the efficiency is dependent on the inductor series resistance we have several efficiencies depending on that series resistance value:
0.1 ohms, 87.5 percent
1.0 ohms, 87 percent
10 ohms, 75 percent

So we can get a quick idea of what using a cheap 'choke' for the inductor will do...lower the efficiency by greater than 10 percent. If this is not a concern then no big deal, but that's a lot to me when something has to operate off of a battery because that translates to a longer run time. Since the efficiency is based on power and the input voltage is relatively constant, that means 10 percent longer run time. With an LED maybe not as important as that translates to 10 percent longer run time at a given brightness level. Lets see...10 hours run time would mean one more hour run time with an increase of 10 percent efficiency. Also, transistor would stay cooler if it got warm to begin with at this lower current level. Obviously higher current outputs would not be as good unless the drive transistor was replaced with a very low sat type.
Keep in mind i was testing with the 2N4401 and 2N4403 types not the BC types given in the original schematics.

 

[Mr RB]

Hi MrAl, sorry my mistake when you said you removed the cap I just assumed you removed C1, the zener cap. If you removed the positive feedback cap C2 it's not going to oscillate well or at all.

If you want to get the efficiency up, there are a couple of things to look at;
1. Switching losses; these can be reduced by running the lowest freq possible as every switching "event" wastes a parcel of power. Of course to run low frequency without inductor saturation you need a good inductor. Making the switching faster helps but the SMPS already switches quite cleanly.
2. ON losses; mainly Q1 saturation loss and current through R1 and R2. If R2 is too high you get worse Q1 sat, if R2 is too low Q1 saturates nicely but you waste power through R1 and R2.
3. OFF losses; mainly the diode saturation, using a 0.3v schottky diode will be much better than a 0.6v diode.
4. Inductor losses; the current is always on through the inductor so I2R losses are inportant. A toroid generally has low ohms so losses are low there, and allows higher core flux so you can run lower frequency and higher current ripple.
5. Voltages, running Vin approx double Vout (or a bit higher) gives approx 50% duty where efficiency is highest. Also, the higher Vin and Vout are the lower the ON and OFF saturation losses are in comparison so a 24v-12v converter will be more efficienct than a 12v-6v converter.

I'm still curious if the Spice sim can get an efficient SMPS happening at 5mA output (like I said in above post)? Maybe with the right inductor choice?