Reverse sawtooth generator schematic needed with discrete components

[neazoi]

I refer to the circuit in post #17.
I have used 2k2 for R2 and 47R for R6. These were the only changes. I do not know why is it so non-linear, any ideas?
For the cap I used 100nF. 30 Hz and lower is perfect for my application but I would like ot fo be linear.

Now about the hot transistor,
Ok I left the R6 connected durring tests on 50R and also 1M. This was not right and that was why the transtistor was probably getting hot. Maybe I should replace it with 1M, to match my scope's 1M input and to limit the current to the transtistor? 50 ohm is not really needed for this application. I do not like to have the internal load of the oscilloscope dissipating all the power of the amplifier so I think I am going to include this R6. But then R6 in parallel to the internal resistance of the scope will do 500K. Is that a problem?

 

[iimagine]

Observing current being charged into C1, I noticed that around 1.2A of current (green line) is being injected into the cap. That is a lot of current going through the poor charging transistor (0.6A). To minimized this, I've added a cap at the base of the charging transistor, slowing the current charging into the cap down to ~5mA. Here are the outputs taken from the original and modified.

The same modification is to be applied to my circuit as well.

- - - Updated - - -

I refer to the circuit in post #17.
I have used 2k2 for R2 and 47R for R6. These were the only changes. I do not know why is it so non-linear, any ideas?
For the cap I used 100nF. 30 Hz and lower is perfect for my application but I would like ot fo be linear.

As I said earlier, linearity can not be achieved using this method of discharging, because the VBE diff of the discharging transistor is constantly changing.

 

[neazoi]

Observing current being charged into C1, I noticed that around 1.2A of current (green line) is being injected into the cap. That is a lot of current going through the poor charging transistor (0.6A). To minimized this, I've added a cap at the base of the charging transistor, slowing the current charging into the cap down to ~5mA. Here are the outputs taken from the original and modified.

The same modification is to be applied to my circuit as well.

- - - Updated - - -



As I said earlier, linearity can not be achieved using this method of discharging, because the VBE diff of the discharging transistor is constantly changing.

Thanks I will try this change today and let you know. I do not expect a great linearity, but much better than the one shown in the tested circuit. The NON-reverse sawtooth presented at post #1 is very linear by the way!

 

[iimagine]

The NON-reverse sawtooth presented at post #1 is very linear by the way!

Linearity can only be obtained from a constant current source or sink, therefore the circuit posted and tested by the original author is a true linear ram generator, the bootstrapping method was suggested by someone else, which was the one that you prefer to have it inverted, was not tested.

Anyway, why not try mine?

- - - Updated - - -

I did not notice that you use 100nF cap. Try replacing voltage divider 10k/2.2k with 1k/220, and the cap at the charging transistor base to 1uF

 

[neazoi]

Linearity can only be obtained from a constant current source or sink, therefore the circuit posted and tested by the original author is a true linear ram generator, the bootstrapping method was suggested by someone else, which was the one that you prefer to have it inverted, was not tested.

Anyway, why not try mine?

That is what I said on post #3 but I was proposed for the schematic for the bootstraping version.
Are your circuit based on constant current method?
The fact that I see great linearity in tested conditions, at the first circuit of the author, if your method does not work as expected in practice than how could I convert the first schematic posted on #1 for inverse sawtooth?
(I guess I could use the same gurrent amplifier proposed.)

 

[iimagine]

That is what I said on post #3 but I was proposed for the schematic for the bootstraping version.
Are your circuit based on constant current method?
The fact that I see great linearity in tested conditions, at the first circuit of the author, if your method does not work as expected in practice than how could I convert the first schematic posted on #1 for inverse sawtooth?
(I guess I could use the same gurrent amplifier proposed.)

Yes, my circuit employed a constant current sink
I will try and convert that circuit for you

 

[neazoi]

You mean convert the circuit in post #1? I would really appreciade. Thank you
I would also try yours, I am just worried of the lower amplitude waveform.

 

[crutschow]

I refer to the circuit in post #17.
I have used 2k2 for R2 and 47R for R6. These were the only changes. I do not know why is it so non-linear, any ideas?
For the cap I used 100nF. 30 Hz and lower is perfect for my application but I would like ot fo be linear.

Now about the hot transistor,
Ok I left the R6 connected durring tests on 50R and also 1M. This was not right and that was why the transtistor was probably getting hot. Maybe I should replace it with 1M, to match my scope's 1M input and to limit the current to the transtistor? 50 ohm is not really needed for this application. I do not like to have the internal load of the oscilloscope dissipating all the power of the amplifier so I think I am going to include this R6. But then R6 in parallel to the internal resistance of the scope will do 500K. Is that a problem?

I think the gross non-linearity your are seeing is due to a wiring or part error. Did you double-check your circuit?

As I stated any value for the load (R6 in parallel with any other load) of 200Ω or greater should be fine.

 

[neazoi]

I think the gross non-linearity your are seeing is due to a wiring or part error. Did you double-check your circuit?

As I stated any value for the load (R6 in parallel with any other load) of 200Ω or greater should be fine.

The circuit is correct. Thanks for notifying me about the resistor.

 

[iimagine]

As promised, here is Alen Yate's Linear Ramp Inverted

Again, look at the current being charged into the cap, 600mA! To reduce this, use the same method I described earlier, use a 10uF cap to reduce it down to ~5mA

 

[neazoi]

As promised, here is Alen Yate's Linear Ramp Inverted

Again, look at the current being charged into the cap, 600mA! To reduce this, use the same method I described earlier, use a 10uF cap to reduce it down to ~5mA

Thank you very much!
I will try this probably today and let you know.

 

[iimagine]

Just a reminder if you decided to construct my circuit, the PNP transistor is meant to have its polarity reversed, which means its collector is to be connected to the + supply voltage. If you dont do this, it will not osc.

 

[iimagine]

Well, I decided to test out my own circuit, and here is the result

True linear as expected. Scratched the PNP reversed polarity idea, although it did osc but it was so unstable, it produced periodic waveform appearances (I would have to study more about this, since it's purposes was to minimized the gain produced by both transistors); So I turned it around the proper way and a stabled waveform appeared. At first, the waveform looked distorted, was not linear as expected???! So I double checked everything and tried it again... still non-linear!!!? I got a little confused as how this can be... I have a constant current sink darn it! And it became clear to me, I was suspecting that my scope's probe is loading down the output, so I quickly assembled crutschow's emitter buffer stage and.. Whoop! there it is!

Since reversed polarity didnt work, the waveform appearance is just as what Alen Yate observed, although he claimed that it has to do with voltage exceeding 8V causing transistor BE breakdown, I'm only using 5.75V. If I had a 100nF handy I could have tried it at the base of the charging transistor to see if it would reduce this effect somehow.

 

[crutschow]

Here's a modified immagine circuit for your interest. I added an emitter diode to prevent reverse bias breakdown and added a resistor to bias the current-mirror separately to prevent feedback that was causing the small non-linearity at the top of the waveform.

 

[neazoi]

Here is the waveform from the last circuit without a buffer, as it is shown on the post #30.
I have used 2n2222 and 2n2907 for the transistors.
Not very linear. Also no oscillation fov vcc more than 4v

Shall I try also circuit in #34 to see how it performs if it is more linear?

 

[iimagine]

Here is the waveform from the last circuit without a buffer, as it is shown on the post #30.
I have used 2n2222 and 2n2907 for the transistors.
Not very linear. Also no oscillation fov vcc more than 4v

Which circuit is this? Is this the one that I converted?

Edit: Never mind I just noticed the post number

Hmm,... this is strange, the conversion is the exact opposite of the original

 

[neazoi]

Which circuit is this? Is this the one that I converted?

Please seem my post above, I have edited it.

 

[iimagine]

Shall I try also circuit in #34 to see how it performs if it is more linear?

well, did you look at my test result? the decision is your

- - - Updated - - -

you would have to feed it through a buffer stage to tell if it is linear or not, it does not take that long to assemble another circuit

 

[neazoi]

So many circuits have been posted and it starts to get a bit confusing now. Is the circuit in post #34 the iimagine's final circuit?

 

[crutschow]

So many circuits have been posted and it starts to get a bit confusing now. Is the circuit in post #34 the iimagine's final circuit?

Yes, it is getting a little confusing. :-? The circuit in #34 is includes my modifications to iimagine"s circuit to solve a couple problems I noted and added the output buffer. Note that I also changed some resistor values which seemed to improve circuit performance and increase the output voltage with a 12V supply.