PopularFX
Home Help Search
Advanced search 
Login Register
Welcome,Guest. Please login or register.
2025-04-01, 01:43:26
News: Registration with the OUR forum is by admin approval.

Pages: 1 2 3 4 [5] 6 7 8 9
Author Topic: Tetra Replication  (Read 16381 times)

Group: Professor
Hero Member
*****

Posts: 3670
I have two SDG 1032X signal generators, I was hoping to link them to have it generate a 3 phase signal but it's too limited. Works well as a function generator though.
Maybe you can.  If you create two signals 120º out of phase you can generate the 3rd signal simply by adding and inverting the two.
Alternatively, you can synchronize two SDG 1032X dual channel signal generators using the Aux and 10MHz In/Out connectors on the rear panel and generate up to 4 phase-synchronized signals.
The caveat is that these signal generators are a little too slow for your purposes because their minimal pulse width is 32.6ns and you need them to be around 20ns.

What is the minimal pulse width in the Square Wave mode of you set the Duty Cycle to 0.001% ?  Try this at high and low Square Wave frequencies...

Last but not least, even if you cannot get the pulse width below 32ns, you can still use these signal generators to generate the three 3-phase signals and use only one edge of the pulses (e.g. the rising edges) to trigger a pulse stretcher such as the 74VHC221.

Edit: I'm in the US
I'm in EU so I cannot help you with test equipment but if you are somewhere near NY then perhaps Chet could help you.
« Last Edit: 2025-02-24, 11:57:56 by verpies »
   
Group: Moderator
Sr. Member
*****

Posts: 308
Maybe you can.  If you create two signals 120º out of phase you can generate the 3rd signal simply by subtracting the two.
Alternatively, you can synchronize two SDG 1032X dual channel signal generators using the Aux and 10MHz In/Out connector on the rear panel and generate 4 synchronized signals.
The caveat is that these signal generators are a little too slow for your purposes because their minimal pulse width is 32.6ns and you need them to be around 20ns.

What is the minimal pulse width in the Square Wave mode of you set the Duty Cycle to 0.001% ?  Try this at high and low Square Wave frequencies...

Last but not least, even if you cannot get the pulse width below 32ns, you can still use these signal generators to generate the three 3-phase signals and use only one edge of the pulses (e.g. the rising edges) to trigger a pulse stretcher such as the 74VHC221.
I'm in EU so I cannot help you with test equipment but if you are somewhere near NY then perhaps Chet could help you.

I've tried to get this to work with the signal generators but this model doesn't support it. Even following their guides does not work. I think the model one step up does it, so I just bought the wrong model based on an error in documentation.
   
Group: Moderator
Sr. Member
*****

Posts: 308
I can't quite remember, I will give it another shot. Lowest pulse width is 16.6 ns from this unit.
   

Group: Professor
Hero Member
*****

Posts: 3670
I've tried to get this to work with the signal generators but this model doesn't support it.
The 10MHz clock reference In/Out connectors don't work or generating two sine wave signals 120º out of phase doesn't work ?

If the latter is possible then 1 op-amp and 3 comparators (for example these) will generate 3-phase rectangular pulses several nanoseconds wide, while the PRF is digitally controllable by one SDG 1032X.
   
Group: Moderator
Sr. Member
*****

Posts: 308
The 10MHz clock reference In/Out connectors don't work or generating two sine wave signals 120º out of phase doesn't work ?

If the latter is possible then 1 op-amp and 3 comparators (for example these) will generate 3-phase rectangular pulses several nanoseconds wide, while the PRF is digitally controllable by the SDG 1032X.

Do you have a circuit diagram for that? I've been using inverters and AND gates to make pulses as my go-to.
   

Group: Professor
Hero Member
*****

Posts: 3670
I've been using inverters and AND gates to make pulses as my go-to.
That is an option too ...but does it allow you to digitally control the PRF or the pulse width in the 20-30ns range ?
The circuit you posted generates square pulses - not rectangular ones with variable duty cycle and widths.  In other words, that circuit generates pulses which are as wide as the gaps between them.
   
Group: Moderator
Sr. Member
*****

Posts: 308
That is an option too ...but does it allow you to digitally control the PRF or the pulse width in the 20-30ns range ?
The circuit you posted generates square pulses - not rectangular ones with variable duty cycle and widths.  In other words, that circuit generates pulses which are as wide as the gaps between them.

Yeah I was tinkering around with having one global pulse width control but if the delays aren't correct it causes double pulsing. Here is the current way I'm planning. That buffer after each inverter is going to be a DS1023 chip for 0-255ns delays, so around 15-270ns pulse widths.
   

Group: Professor
Hero Member
*****

Posts: 3670
Do you have a circuit diagram for that?


The op-amp "A" works like this.  The resistors should be equal: R1=R2=R3.
Phase 1 and Phase 2 signals can come from the SDG1032X (120º out of phase) and if you'd like to have the pulse widths adjusted digitally, too, then substituting P1 with one of these cheap Chinese digital voltage regulators would give you that feature.  (there are even cheaper ones out there - I just did not take the time to look.  Once, when browsing, I have seen a digital voltage regulator for ~$5. It was weak (100ma) but that is more than enough for this purpose.)

You can also control the output pulse widths by controlling the amplitudes of the Phase 1 and Phase 2 input signals but they both must be changed by the same amount (I don't know whether the SDG1032X can control the amplitudes of both of its channels together).  It is important to mention that changing the amplitudes of the Phase 1 and Phase 2 signals DOES NOT affect the amplitudes of the 3 output pulses (only their width is affected).
« Last Edit: 2025-02-24, 11:37:55 by verpies »
   
Group: Moderator
Sr. Member
*****

Posts: 308


The op-amp "A" works like this.
Phase 1 and Phase 2 signals can come from the SDG1032X (120º out of phase) and if you'd like to have the pulse widths adjusted digitally, too, then substituting P1 with one of these cheap Chinese digital voltage regulators would give you that feature.  (there are cheaper ones out there - I just did not take the time to look.  Once, when browsing, I have seen a digital voltage regulator for ~$5. It was weak (100ma) but that is more than enough for this purpose.)

Looks good, I'll run it thru simulation! Op amps tend to be slow so I never consider them, but perhaps I should have. Looks like a great way to fill in for the missing phase.
   
Sr. Member
****

Posts: 280
Another option to throw into the mix would be to generate PWM signals using a microcontroller.

I'm using a Teensy 4.1 with PlatformIO to generate PWM signals. It specifically supports generating 3 phase synchronised signals. It also supports dead time insertion and a host of other features. Page 3091 of the NXP processor reference manual has extensive documentation on what is supported.

eFlexPWM is a nice library which makes it easier.

I've created a rudimentary web UI to allow adjustment of PWM parameters on the fly. I'm also logging various metrics to an InfluxDB time-series database, which allows you to create nice graphs and charts to visualise the data.
   

Group: Professor
Hero Member
*****

Posts: 3670
Another option to throw into the mix would be to generate PWM signals using a microcontroller.

I'm using a Teensy 4.1 with PlatformIO to generate PWM signals. It specifically supports generating 3 phase synchronised signals.
What is the minimum pulse width that you can generate this way ?
   
Sr. Member
****

Posts: 280
What is the minimum pulse width you can generate this way ?

The minimum pulse width looks to be about 6ns with stock speeds.
   

Group: Professor
Hero Member
*****

Posts: 3670
The minimum pulse width looks to be about 6ns with stock speeds.
That is the 1 clock tick period of the PWM block, isn't ?
BTE wants a high PRF (for example: 20ns ON, 300ns OFF) and 3 such signals 120º apart.
With what granularity can these pulse widths be varied by the Teensy 4.1?
   
Group: Moderator
Sr. Member
*****

Posts: 308
Another option to throw into the mix would be to generate PWM signals using a microcontroller.

I'm using a Teensy 4.1 with PlatformIO to generate PWM signals. It specifically supports generating 3 phase synchronised signals. It also supports dead time insertion and a host of other features. Page 3091 of the NXP processor reference manual has extensive documentation on what is supported.

eFlexPWM is a nice library which makes it easier.

I've created a rudimentary web UI to allow adjustment of PWM parameters on the fly. I'm also logging various metrics to an InfluxDB time-series database, which allows you to create nice graphs and charts to visualise the data.

I love a good digital data acquisition setup...soon...

Going to investigate various timing options for pulse width control. Right now ds1023 looks good.

Pwm digitally is really nice too. I'd love a computer controlled setup for automatic sweeping of parameters.
   

Group: Professor
Hero Member
*****

Posts: 3670
Op amps tend to be slow so I never consider them, ...
Nowadays, there are 1GHz op-amps for $8.
They get slower when they amplify signals more than x1, but at the unity gain their bandwidth remains the same as in their datasheets.
This circuit operates at unity gain when R1=R2=R3.
   
Sr. Member
****

Posts: 280
BTE wants a high PRF (for example: 20ns ON, 333ns OFF) and 3 such signals 120º apart.
With what granularity can these pulse widths be varied by the Teensy 4.1?

A 353ns period is 2,832,861 Hz which is no problem for the Teensy.

You can adjust the period widths with 6ns precision, so considering this example, you could vary it from 6ns to 347ns.

Operating at a frequency of 2,832,861 Hz with a 150MHz bus clock (from which the PWM signal is derived), there are 52.95 ticks in a PWM period. There are 0.15 clock ticks per nanosecond, so that means the minimum resolution is 6.67ns.

Of course you can overclock the CPU and this will also overclock the bus, since the bus clock is CPU clock / 4. If you overclocked it to 1GHz then you would have a bus clock of 250MHz which would increase the resolution to 4ns.

With a 4ns resolution you would be able to adjust the PWM signal like so:

OnOff
4349
8345
12341
16337
20333
24329

and so on.
   

Group: Professor
Hero Member
*****

Posts: 3670
With a 4ns resolution you would be able to adjust the PWM signal like so:

OnOff
4349
8345
12341
16337
20333
24329

and so on.

@BTE:  Is ..., 20ns, 24ns, 28ns, 32ns, 36ns, 40ns, ... pulse width adjustment granularity sufficient for your purposes ?
   
Group: Moderator
Sr. Member
*****

Posts: 308
Yes
   

Group: Professor
Hero Member
*****

Posts: 3670
Yes

What about this ?



The target is only 50-160kHz wide.
   
Group: Moderator
Sr. Member
*****

Posts: 308
What about this ?


The first tests will have a magnetic field around the experiment to lower the pulse rate. Then I will gradually lower the field strength and try to plot frequency vs field strength to get a starting point in regards to required pulse rate
   

Group: Professor
Hero Member
*****

Posts: 3670
The first tests will have a magnetic field around the experiment to lower the pulse rate. Then I will gradually lower the field strength and try to plot frequency vs field strength to get a starting point in regards to required pulse rate

The change of the pulse width from 20ns to 24ns constitutes -17% change of the fundamental frequency component.  That is much more than you can achieve by varying the external magnetic field at home (-0.033%).

20ns → 50.0MHz
24ns → 41.7MHz
28ns → 35.7MHz
32ns → 31.25MHz
36ns → 27.8MHz
40ns → 25.0MHz

According to this graph, the target is only 50-160kHz wide.
« Last Edit: 2025-02-24, 14:44:06 by bte »
   
Group: Moderator
Sr. Member
*****

Posts: 308
Interesting. I didn't know pulse width changes fundamental frequency!
   

Group: Professor
Hero Member
*****

Posts: 3670
Interesting. I didn't know pulse width changes fundamental frequency!
That's basic spectral analysis.
The pulse width affects the frequency of the strongest component of the spectrum and which frequencies are absent from the spectrum while the PRF affects the spectrum density of the remaining components, as I illustrated here.

You can read about the details here and here.

« Last Edit: 2025-02-25, 01:46:11 by verpies »
   
Group: Moderator
Sr. Member
*****

Posts: 308
Does that alter the fundamental or just vary the power of harmonics?
   
Group: Moderator
Sr. Member
*****

Posts: 308
   
Pages: 1 2 3 4 [5] 6 7 8 9
« previous next »


 

Home Help Search Login Register
Theme © PopularFX | Based on PFX Ideas! | Scripts from iScript4u 2025-04-01, 01:43:26
Loading...