12V - 350V 200mA converter for motorcycle CDI

[abuhafss]

Whilst you still have relatively easy access to it, it's a good idea to compare the performance with and without the gap. With all other things being equal, I'd go for the gapped version.

With a gap of 0.6mm the transformer is super quite but, what about the increase in the temperature. Is it normal for a ferrite core transformer?

 

[(*steve*)]

The increase in temperature normally comes from the windings.

If it is combined with lower output power and heating (often excessively) of the driver transistors it could be a symptom of the inductor saturating.

If it is combined with higher output power it is more likely to be cause by resistive losses as part of the normal operation. In this case it may indicate that you need to reduce the power losses -- this may mean using thicker wire in the transformer.

It is normal for transformers to become warm in operation. But they should never get so hot that you would feel at any risk if you held your finger indefinitely on them (i.e. warm, not hot). From somewhere in the dark recesses of my mind, a 20ºC rise over ambient is something like what you should be expecting when operating a transformer close to its rated power (internally it will be significantly hotter).

 

[(*steve*)]

Now, the transformer was quite such that I doubted it was not working at all. I checked the oscillation frequency at the base of Q1, it was 248 kHz. However, I noticed that this frequency decreased gradually. After about 5 minutes, it was 240kHz and then later it was 236kHz. I also noticed an increase in the temperature of the transformer.

None of those things is unusual. 248kHz is a very high frequency, you will have to ensure your diodes are fast. I assume the ones Kris and others have specified are fast.

Here are the Collector Voltage Waveforms:

The most important thing is the vertical scale. How many volts per division? Other issues might be: x10 probe? AC or DC coupled? Where is 0 volts?

During bench test with 13.5V/2A (without ignition coil), the TIP41 remain cool.
During test run with the two 3.3k/1W resistors connected, some fuse went off. After changing the fuse I removed the resistors and it worked normally. However, the TIP41 showed no improvement in heating..............asking for a heat-sink in just 3 minutes.

Did it ask using smoke signals, or just by feeling hot? If the latter, then just add a heatsink.

An alternative is to use a mosfet which may have much lower losses.

R1 was warm and the newly replaced R2 330Ω/1W was cool. Also the original C8 stopped working sometime during the bench test.

There are 2 C8's marked on the circuit. I assume you mean the one at the output?

It needs to be rated for both high voltage and high current. I imagine that you'd need to pick a part very carefully for long life in this position.

 

[abuhafss]

None of those things is unusual. 248kHz is a very high frequency, you will have to ensure your diodes are fast. I assume the ones Kris and others have specified are fast.

Yes, they are fast diodes.

The most important thing is the vertical scale. How many volts per division? Other issues might be: x10 probe? AC or DC coupled? Where is 0 volts?

Probe at x10, AC coupled
DC Voltage measured at the collector using multimeter, it was 11.86V.
Volts/div & 0 volts=> Please bear with me, I am not an experienced user of scope. I shall get the reading again.

Did it ask using smoke signals, or just by feeling hot? If the latter, then just add a heatsink.

An alternative is to use a mosfet which may have much lower losses.

As stated earlier, the original Yamaha OEM circuit has no heatsink. And normally, the regular CDIs (not the performance CDIs) do not have heatsink; the entire PCB+components are covered/sealed with epoxy sealant.

Although the original circuit has TIP****, a mosfet can be considered if some cost effective device is available.

There are 2 C8's marked on the circuit. I assume you mean the one at the output?

Oh sorry. Yes, I was referring to the 1μF/500V capacitor.

It needs to be rated for both high voltage and high current. I imagine that you'd need to pick a part very carefully for long life in this position.

Shall appreciate, if you could elaborate please.

I had used something like this

http://i01.i.aliimg.com/wsphoto/v0/...400v-b-font-metallized-polypropylene-film.jpg

 

[(*steve*)]

Volts/div & 0 volts=> Please bear with me, I am not an experienced user of scope. I shall get the reading again.

That's critical. And if you use DC coupling and you tell us where zero is, we can see exactly what the voltages are.

As stated earlier, the original Yamaha OEM circuit has no heatsink. And normally, the regular CDIs (not the performance CDIs) do not have heatsink; the entire PCB+components are covered/sealed with epoxy sealant.

Although the original circuit has TIP****, a mosfet can be considered if some cost effective device is available.

I think the issue may be the frequency this is operating at. Although the transistor you're using has an F_T of 3MHz, that's at a collector current of 500mA, and I suspect you've got a higher collector current. Also you may be pushing the transistor into saturation which has the double edged sword of lowering ON dissipation, but slowing the transition from On to Off.

I'm going back to review those oscilloscope traces again...

OK, I'm not certain where 0V is, but I'm seeing big spikes in both directions. I hope we're not seeing reverse currents. Is this a dual trace oscilloscope? I'd like to see base and collector voltage with respect to earth.

Rechecked the transistor. you have a TIP31, not a TIP41, but they're similar Problem is that TIP41 has a breakdown of 40V, and that could easily be exceeded. A, B, and C variants have higher breakdown voltages. Is this a plain TIP31? The breakdown will damage the transistor, but it will also cause significant heating because you have both high current and a large voltage drop, even though the duration is small.

Oh sorry. Yes, I was referring to the 1μF/500V capacitor.

Shall appreciate, if you could elaborate please.

I had used something like this

http://i01.i.aliimg.com/wsphoto/v0/...400v-b-font-metallized-polypropylene-film.jpg

I'm no expert here. Is this is what is used in other CDI applications?

What it looks like doesn't help much. I'd have to look at the specs, but since I'm not sure exactly what I'd be specifically looking for, I'd look for recommendations about capacitors for this particular application.

Metalised polypropylene caps are generally "self healing", which means that breakdowns in the dielectric tend to vaporise the metal causing an open rather than a short. This reduces the capacitance a little.

How did it fail? short, exploded, something else?

 

[abuhafss]

That's critical. And if you use DC coupling and you tell us where zero is, we can see exactly what the voltages are.

So do you want me to use DC coupling and then take waveforms?

I think the issue may be the frequency this is operating at. Although the transistor you're using has an F_T of 3MHz, that's at a collector current of 500mA, and I suspect you've got a higher collector current. Also you may be pushing the transistor into saturation which has the double edged sword of lowering ON dissipation, but slowing the transition from On to Off.

I am also suspecting the frequency. May I just remind a point which give us some clue.......the original circuit is using unidirectional TVS 200V diode whereas, I am using 4 x 82 zeners. Which means the original circuit is supposed to regulate at 200V and I intended to have the regulation at 300-330V. Can this regulation voltage have any sort of impact on the operation frequency?

Is this a dual trace oscilloscope? I'd like to see base and collector voltage with respect to earth.

Okay, I shall get those traces. You need them with AC coupled or DC coupled?

Rechecked the transistor. you have a TIP31, not a TIP41, but they're similar Problem is that TIP41 has a breakdown of 40V, and that could easily be exceeded. A, B, and C variants have higher breakdown voltages. Is this a plain TIP31? The breakdown will damage the transistor, but it will also cause significant heating because you have both high current and a large voltage drop, even though the duration is small.

Hmm.....that could be the main cause. But, as far as I remember it should be TIP41C (I never had TIP41 in my collection). Unfortunately, I could not go to my workshop today. I shall check it tomorrow. BTW, I also have some KSD2058 (3A) with me but didn't used them because of low F_T of 0.4MHz.

I'm no expert here. Is this is what is used in other CDI applications?

What it looks like doesn't help much. I'd have to look at the specs, but since I'm not sure exactly what I'd be specifically looking for, I'd look for recommendations about capacitors for this particular application.

Metalised polypropylene caps are generally "self healing", which means that breakdowns in the dielectric tend to vaporise the metal causing an open rather than a short. This reduces the capacitance a little.

Yes, usually CDIs have similar bulky capacitors rated 400V or more. However, recently I came across smaller versions of metalized polypropylene caps rated 500V. I have used them in CDIs for test runs and found them to be working ok but, not sure if they would perform normally in the long run.

How did it fail? short, exploded, something else?

Something else; I just noticed it not working during test run so I replaced it.

 

[(*steve*)]

Use DC coupled. And note on your images where zero is (i.e. where the trace is when the probe is connected to ground. Also note the probe type (presumably x10) and the vertical sensitivity (volts/division) and the horizontal timebase (μs/div). Make sure the divisions are visible, and also that the traces are separated enough so we can't confuse one with the other (minimise overlay if possible)

 

[abuhafss]

First of all, Q1 is TIP41C.

Okay now, here are the traces. Both obtained with DC coupling and probes at x10. The upper one is for the VBASE and the lower is for VCOLLECTOR.





Test was carried out at bench without the ignition coil with 2 x 3.3k/1W resistors in series connected across D8.
Current drawn was 1.60A and oscillation frequency was read 200kHz. With the passage of time the current drawn and the oscillation frequency increased. The transformer also gained temperature.

 

[(*steve*)]

Are you sure the base voltage was at 20mV/Division? This seems to suggest the transistor is turning on with less than 400mV of base voltage. 50mV/division seems more likely.

I'm not seeing any voltage spikes on the collector voltage (and that's good), but the traces seem to indicate you're running from 24V (or is it set for 0.5V/division?). Oh, and was zero set to the bottom gratucule, or slightly below? -- it's best to set it on a graticule.

To test the vertical scale, measuring the 12V battery on the DC scale with the vertical amp set at 1V/div and with a x10 probe, the beam should jump 1.2 divisions up. We're seeing more than 2 here with no real indication that it's due to an inductive spike (however it could be -- even though the shape seems too "nice")

It would also be interesting if you can add that small resistor in series with the collector and show the voltage between ground and the collector, and between ground and the top of the resistor.

Again, both channels set to DC input, and if you can get your scope to show A-B (channel A should be the top of the resistor, channel B the collector).

If you're not sure about the A-B thing, show both traces and I can figure it out. I'm looking to see if there is current flowing through the transistor when it's off.

 

[KrisBlueNZ]

Sorry for dropping out of the thread! Steve is doing a good job though.

I really want to see a primary current waveform! The way you did it before, with a series resistor between the transistor's collector and the bottom of the primary, is not a good idea, because there are high-voltage spikes at that point in the circuit. Your oscilloscope's ground is probably connected to mains earth. Even if your circuit is electricaly isolated from mains earth (and if it's not, connecting the scope earth to the collector would have raised merry hell), there is still a lot of stray capacitance between the rest of the circuit and the earth, and this will cause the scope to indicate voltages that aren't really there.

You could try a low-value (e.g. 0.1Ω or 0.47Ω) resistor in series with the emitter, and connect the scope across that. Otherwise, connect it at the top of the primary and hook the scope with its ground lead to the positive supply and the signal lead to the primary side of the resistor. This will make the current display upside-down but at least most of the common-mode noise will be eliminated. The resistor must be non-inductive - don't use a wirewound resistor unless it's specially made to be non-inductive.

I'm pretty concerned about the frequency though. 200 kHz seems awfully high. But this circuit depends very much on the characteristics of the transformer, including its saturation characteristics. That is, if it's a blocking oscillator. Steve, do you agree that it is a blocking oscillator?

Also as Steve pointed out, you need a transistor with a higher collector-emitter voltage rating than 40V. A TIP31C or TIP41C. But those are not fast transistors. I think we need to figure out why the oscillator is running at such a high frequency.

 

[KrisBlueNZ]

One other thing. Can you add a good-quality 1 µF 50V capacitor e.g. a film cap, as directly as possible between the top side of the primary and the emitter of the transistor. Just to give the transistor a nice clean low-impedance source to work into. Something like http://www.digikey.com/product-detail/en/B32522C1105K289/495-4895-1-ND/3881088

Also can you upload a photo of your construction.

 

[abuhafss]

Are you sure the base voltage was at 20mV/Division? This seems to suggest the transistor is turning on with less than 400mV of base voltage. 50mV/division seems more likely.

Yes, it was at 20mV/DIV -- positively.

I'm not seeing any voltage spikes on the collector voltage (and that's good), but the traces seem to indicate you're running from 24V (or is it set for 0.5V/division?). Oh, and was zero set to the bottom gratucule, or slightly below? -- it's best to set it on a graticule.

It was set 1V/Div and zero was set exactly to the bottom graticule (not slightly below).

It would also be interesting if you can add that small resistor in series with the collector and show the voltage between ground and the collector, and between ground and the top of the resistor.

Do you mean, I add something like 0.22Ω between collector and the primary of transformer?

Again, both channels set to DC input, and if you can get your scope to show A-B (channel A should be the top of the resistor, channel B the collector).

If you're not sure about the A-B thing, show both traces and I can figure it out. I'm looking to see if there is current flowing through the transistor when it's off.

This scope practical training is more prized than watching scope tutorials

 

[abuhafss]

Sorry for dropping out of the thread!

I assumed you have been preoccupied. Today, I looked for you and your recent post revealed your presence. Anyway, welcome back .

You could try a low-value (e.g. 0.1Ω or 0.47Ω) resistor in series with the emitter, and connect the scope across that.

Okay, I'll do that with a ceramic resistor.

I'm pretty concerned about the frequency though. 200 kHz seems awfully high.

Today, I mistakenly left the circuit powered on and about an hour later the frequency had climbed to 500kHz !

Steve pointed out, you need a transistor with a higher collector-emitter voltage rating than 40V. A TIP31C or TIP41C. But those are not fast transistors. I think we need to figure out why the oscillator is running at such a high frequency.

Agreed, and just to remind you the OEM circuit had TIP****....could it be a darlington?

 

[abuhafss]

One other thing. Can you add a good-quality 1 µF 50V capacitor e.g. a film cap, as directly as possible between the top side of the primary and the emitter of the transistor. Just to give the transistor a nice clean low-impedance source to work into. Something like http://www.digikey.com/product-detail/en/B32522C1105K289/495-4895-1-ND/3881088

Yes, I can do that for testing. But for actual, no space
And we are going away from the original design.

Also can you upload a photo of your construction.

Yes, I'll do that tomorrow morning.

 

[abuhafss]

To test the vertical scale, measuring the 12V battery on the DC scale with the vertical amp set at 1V/div and with a x10 probe, the beam should jump 1.2 divisions up. We're seeing more than 2 here with no real indication that it's due to an inductive spike (however it could be -- even though the shape seems too "nice").

Could it be a calibrating issue of the scope?
I acquired this scope last month and don't know when it was last calibrated.

 

[KrisBlueNZ]

Today, I mistakenly left the circuit powered on and about an hour later the frequency had climbed to 500kHz !

And it's only the transformer that's getting hot?

Agreed, and just to remind you the OEM circuit had TIP****....could it be a darlington?

Could be, but I doubt it. There's a lot of base drive available from the transformer. I think it's more likely a standard BJT. Do you still have the original transistor? Does it still work? You can measure the base-emitter voltage using a multimeter on diode test range. That will tell you whether it's a Darlington or not.

 

[abuhafss]

And it's only the transformer that's getting hot?

No, TIP41C melted the tip of scope's probe connected to the collector tap

Could be, but I doubt it. There's a lot of base drive available from the transformer. I think it's more likely a standard BJT. Do you still have the original transistor? Does it still work? You can measure the base-emitter voltage using a multimeter on diode test range. That will tell you whether it's a Darlington or not.

Yeh, I still have it. Will check if it works or not.

 

[abuhafss]

To test the vertical scale, measuring the 12V battery on the DC scale with the vertical amp set at 1V/div and with a x10 probe, the beam should jump 1.2 divisions up. We're seeing more than 2 here with no real indication that it's due to an inductive spike (however it could be -- even though the shape seems too "nice")

Here is the Collector Voltage trace using LT Spice. The waveform is similar to the one we obtained thru our test. The frequency is about 84kHz. The voltage climbs to 43V until interrupted by the trigger pulse.

Oh, and was zero set to the bottom gratucule, or slightly below?

And for reference, here is BASE voltage trace. You can see the trace is going slightly below 0V.



I have attached the files, please change the extensions to CDI DC Yamaha.ASC and BT151.LIB

You may find the simulation helpful to optimize the physical circuit

 

[(*steve*)]

No, TIP41C melted the tip of scope's probe connected to the collector tap

I'll try to put this in a way that won't be easily misinterpreted.

CONNECT A @#$*^&* HEATSINK TO IT!!!​

(Is that too subtle?)

I know you say that the original design doesn't need one, but yours does right now. It's a bit like saying "I'm bleeding out through a gash in my leg, but I refuse to treat it because the original leg didn't have one."

The simulations suggest that I was right about the incorrect vertical amplifier setting for your actual base voltage (or that something else is causing the error)

Voltage seen on the collector during the simulation suggests either the same thing *or* that the transistor is breaking down between emitter and collector.

My suggestion of measuring the collector current using A - B mode on the scope was in an effort to cancel out common mode noise by making what is as close as possible to a differential measurement on your scope (this is mostly in response to Kris). What I would be looking for is the ramped current as the transistor is turned on, and the lack of any forward or reverse spiked currents. The former could be indicative of saturation, the latter of transistor breakdown. Either of these could cause the heating of the transistor. There is some evidence of this from an earlier image, but I need to see it with a DC input so I can be exactly sure where zero current is.

The fact that your simulation shows the collector voltage exceeding the maximum for this transistor is a pretty good indication that you should pick a transistor with a higher V_CEO.

And yes (Kris) it does look like a variant of a blocking oscillator.

 

[abuhafss]

It would also be interesting if you can add that small resistor in series with the collector and show the voltage between ground and the collector, and between ground and the top of the resistor.

Would you please clarify.......do you want me to add 0.22Ω between collector and the primary of transformer?

Actually, I am bit confused over Kris's comments:

The way you did it before, with a series resistor between the transistor's collector and the bottom of the primary, is not a good idea, because there are high-voltage spikes at that point in the circuit. Your oscilloscope's ground is probably connected to mains earth. Even if your circuit is electricaly isolated from mains earth (and if it's not, connecting the scope earth to the collector would have raised merry hell), there is still a lot of stray capacitance between the rest of the circuit and the earth, and this will cause the scope to indicate voltages that aren't really there.