12V - 350V 200mA converter for motorcycle CDI

[KrisBlueNZ]

Steve and I both want to see the current waveform in the transformer primary but we are suggesting different ways of measuring it.

In both cases you need to insert a fairly low-value resistor in series with the primary current path and measure the voltage across that resistor.

Steve's suggestion is to connect the resistor between the bottom of the primary and the collector of the transistor, and use two oscilloscope probes, one at each end of the resistor, with the oscilloscope set to "subtract" mode, where it subtracts the voltage from one probe from the voltage from the other probe. In theory the scope will then display the voltage across the resistor.

The problem with that method is that there's a lot of fast, high-voltage common-mode signal at that point - that is, signal that appears on both ends of the resistor, because the transistor is switching - and unless the oscilloscope's inputs are exactly matched, you won't be able to isolate the small voltage that appears across the resistor, from the other signals that the scope will pick up. I'm not sure that this will be a problem, but I strongly suspect that it will. I guess it may depend on the quality of your oscilloscope.

My suggestion is to put the resistor at the top end of the primary, which is pretty much a stable voltage. Then you connect a single scope probe directly across that resistor. Specifically, you connect the scope's earth to the top end of the resistor (at the power supply) and the probe tip to the bottom end of the resistor, that connects to the primary.

With this arrangement there is no high-voltage signal on the scope probe to cause problems. But the current waveform will be upside down, so use the "invert" switch if your scope has one. This connection also requires that either your scope, or your circuit, be isolated from mains earth, otherwise your circuit's 0V rail will be connected, via mains earth, to the scope probe's earth clip, and when you connect the scope probe earth clip to the positive rail of your circuit, you will short out your power supply.

Steve feel free to correct me on any of this.

 

[abuhafss]

Kris, thanks for the clarification.

I think, should get the traces in both ways.

 

[KrisBlueNZ]

Sure, try both ways

 

[(*steve*)]

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

Yes, and then measure between ground and the collector with one channel, and ground and the other end of the resistor with the other channel, then set the scope to A-B mode to display the difference between these signals.

This method should eliminate much of the noise that Kris talks about because it will be common to both channels.

edit: But if you can try it both ways, you should see which method gives the cleanest signal.

 

[abuhafss]

Yes, and then measure between ground and the collector with one channel, and ground and the other end of the resistor with the other channel, then set the scope to A-B mode to display the difference between these signals.

This method should eliminate much of the noise that Kris talks about because it will be common to both channels.

Here are the traces

 

[abuhafss]

My suggestion is to put the resistor at the top end of the primary, which is pretty much a stable voltage. Then you connect a single scope probe directly across that resistor. Specifically, you connect the scope's earth to the top end of the resistor (at the power supply) and the probe tip to the bottom end of the resistor, that connects to the primary.

Here is the trace obtained thru your configuration

 

[abuhafss]

Today, I salvaged a TIP31C from an old veroboard....found it in good health so decided to test it the above circuit.

It got hot quicker than TIP41C!

The traces as per Steve's configuration, with TIP31C were almost same as posted for TIP41C at #85.

But for Kris' configuration, it was quite different.....here it is

 

[abuhafss]

Also can you upload a photo of your construction.

Here is the photo of my construction (extremely sorry for poor quality).


A few points of observed:

R1 = 100Ω/1W is also getting fairly warm
R2 was 330Ω/0.25W in OEM circuit but here it was roasted so changed to 1W but still it gets quite warm (cannot touch for 1 sec).

And just for your reference, here is HV inverter schematic of Suzuki OEM CDI. Other portions of CDI can be found [URL='http://mastercircuits.blogspot.com/2011/06/suzuki-shogun-oem-dc-cdi.html']here.[/URL]

[URL='http://mastercircuits.blogspot.com/2011/06/suzuki-shogun-oem-dc-cdi.html'] [/URL]
It is almost identical to the Yamaha version, except that the feedback secondary network at the base of the power transistor is bit different. However, when simulated the results are same as that of Yamaha CDI, on which we are working.

 

[(*steve*)]

Have you noticed the zener diode between the collector and base of the 2SC4837? It's there to prevent the destruction of the transistor from high voltage pulses.

Also, did you change your scope to A-B after displaying the two waveforms I suggested?

 

[KrisBlueNZ]

(Oops - I wrote this earlier but didn't click Send. Steve, you've already asked the question, but I've tried to explain it in more detail.)

Re the traces in post #85, you need to set the oscilloscope to differential mode, so it displays the difference between the two input signals. This may be called "A-B" or "B-A" or "DIFF" or "SUBTRACT" or something like that. It's probably selected by the same switch that selects between single trace and dual trace operation. Then you can increase the sensitivity, and you may be able to see the current ramp that you see in post #86.

The trace in post #86 looks right. I think the big jagged jumps aren't really there; they're just caused by common-mode voltage spikes, but the current ramp looks good. There is no obvious levelling-out that would indicate saturation, but I think it must be occurring rapidly at the end of the ramp.

Steve, could he add small extra winding on the transformer and use that to pick up the magnetic flux waveform if he connected a current shunt resistor across it? That would avoid all the common mode crap if it would work.

 

[abuhafss]

Have you noticed the zener diode between the collector and base of the 2SC4837? It's there to prevent the destruction of the transistor from high voltage pulses.

Yes, and as pointed above the feedback network in our circuit is getting fairly hot. Does it mean that feedback network in the Suzuki OEM circuit is better than the Yamaha OEM circuit?

 

[abuhafss]

Also, did you change your scope to A-B after displaying the two waveforms I suggested?

Sorry Steve, I didn't do that coz I misunderstood.

but I've tried to explain it in more detail.

Thanks, it really helped.

Re the traces in post #85, you need to set the oscilloscope to differential mode, so it displays the difference between the two input signals. This may be called "A-B" or "B-A" or "DIFF" or "SUBTRACT" or something like that.

Got it.

It's probably selected by the same switch that selects between single trace and dual trace operation.

The switch you are talking about is indicated with green arrow but, it only has "CHOP" and "ADD" (no "DIFF" or "SUBTRACT").



The "X-Y" indicated by yellow arrow is probably for plotting trace CH1 (x-axis) with respect to CH2 (y-axis).

....add small extra winding on the transformer and use that to pick up the magnetic flux waveform if he connected a current shunt resistor across it? That would avoid all the common mode crap if it would work.

How about altering the number of turns for the feedback secondary .... increase or decrease?

 

[(*steve*)]

OK, the X-Y on the timebase is for something else.

You need to set the mode to Add, then pull the position knob on channel 2 to invert it. you thus have A + (-B), or A - B

In this case, because toy want the voltage on the collector subtracted from the voltage at the base of the coil, make channel A the base of the coil, and channel B the collector. The output will now be a single trace which corresponds to the voltage drop across the resistor (and hence the current through the inductor, resistor, and transistor).

We're looking for either positive spikes or negative transients. Ideally you'll see the current ramp up as the transistor is on, then fall to zero as the transistor turns off. But let's see what you do get.

 

[abuhafss]

OK, the X-Y on the timebase is for something else.

You need to set the mode to Add, then pull the position knob on channel 2 to invert it. you thus have A + (-B), or A - B

In this case, because toy want the voltage on the collector subtracted from the voltage at the base of the coil, make channel A the base of the coil, and channel B the collector. The output will now be a single trace which corresponds to the voltage drop across the resistor (and hence the current through the inductor, resistor, and transistor).

We're looking for either positive spikes or negative transients. Ideally you'll see the current ramp up as the transistor is on, then fall to zero as the transistor turns off. But let's see what you do get.

Got it Steve, Thanks.
Shall revert with the traces, soon.

 

[abuhafss]

Here are the traces:

CH1 connected to the base of primary



CH 2 connected to the Collector and inverted



CH1 + CH2



Today I bought a fresh TIP42C and did the analysis, here are results

CH1, connected to the base of primary (Voltage/Div was reduced)



CH2 connected to the collector and inverted (reduced voltage/div)



CH1 + CH2 (with new TIP42C)



I could not magnify the height of the CH1+CH2 trace. I don't see any ramp, did I made the observations correctly?

 

[(*steve*)]

Yes it looks like you have done that right.

it looks like you're getting a constant current which is weird.

can you tell me the vertical scale used for these measurements and the value of that resistor so I can estimate the current?

 

[abuhafss]

Yes it looks like you have done that right.

it looks like you're getting a constant current which is weird.

can you tell me the vertical scale used for these measurements and the value of that resistor so I can estimate the current?

The one with old TIP42C was at 2V/div x10 and with new TIP42C it was at 1V/div x10.
Resistor was 0.22Ω

 

[KrisBlueNZ]

The current ramp shows pretty clearly in the first photo in post #87.

If you want a better result using the two probe method, you could connect the resistor at the top end of the primary, instead of the bottom end, and use AC coupling on the inputs. This will allow you to increase the gain of both channels and get a larger and more accuate trace.

 

[abuhafss]

The current ramp shows pretty clearly in the first photo in post #87.

Perhaps, you are talking about post #86.

If you want a better result using the two probe method, you could connect the resistor at the top end of the primary, instead of the bottom end, and use AC coupling on the inputs. This will allow you to increase the gain of both channels and get a larger and more accurate trace.

Do you mean that the current is not constant, as Steve said? And using this new method, the ramp will be revealed?
If so then what is causing the transistor to heat up?

 

[KrisBlueNZ]

Yes, sorry, I meant post #86.

Yes, I don't believe the current is constant. I just think you need a better method of measuring it. Either the method you used in post #86 to get the ramp, or the method I just described in post #98. If we're lucky, using the method in post #98 we can avoid the big spikes that are visible in post #86.

There are several possible reasons why the transistor is heating up. Steve and I have listed the ones we could think of. We need to see the switching waveforms so we can figure out which one it is.

As well as the primary current, we need a clear and accurate picture of the base drive voltage. The last one we have is in post #68. It look wrong. But since then you've replaced the transistor. Can you take another picture? You won't be able to show it alongside the primary current waveform if you use the two probe method to measure that, but we don't really need to see them both in the same display.

When you connect your scope to measure the base-emitter voltage waveform, try disconnecting the probe from the base and connecting it to the emitter. Since the probe is shorted to the ground connection, theoretically you shouldn't see any signal. But if you do, there is common-mode noise being generated by the converter, and it will affect the waveform you see when you move the probe to the base.

You may be able to reduce this effect by using a thick and fairly short wire from the 0V rail of the circuit to the earth terminal on the front panel of the scope. Also, what power source are you using for the converter? If it's a power supply, does it have an earth lift option? Can you try a SLA/VRLA battery as the power source? (Not a car battery - they're too dangerous!)