Signal Generator Limits: Siglent SDG1025, Short pulse

[NivagSwerdna]

Hi

I've just taken delivery of my new toy.... A Siglent SDG1025 function generator (25Mhz)...

and now I'm really confused!

(1) I tried to generate a 25MHz Square wave.... it doesn't look at all square!  Is that because I would need a much higher spec to generate such a signal?

(2) I tried to generate a 1uS pulse with 5V hi, 0V low at a period of 1Hz.  It won't let me do this... it says the minimum pulse width is 1mS!  That's 1000x larger than what I wanted!

I assume the above is user error!

Please provide hints

Thanks

[ataradov]

(1) I tried to generate a 25MHz Square wave.... it doesn't look at all square!  Is that because I would need a much higher spec to generate such a signal?

How do you measure it? At frequencies like this, impedance matching is a big deal.

(2) I tried to generate a 1uS pulse with 5V hi, 0V low at a period of 1Hz.  It won't let me do this... it says the minimum pulse width is 1mS!  That's 1000x larger than what I wanted!

That's the limitation of the generator. You've got the cheapest thing on the market, it will have limitations like this.

Although 1 ms is kind of big, it should be able to do much better that this.

[NivagSwerdna]

How do you measure it? At frequencies like this, impedance matching is a big deal.

Into my pimped up (100MHz modded) Rigol 1052E scope with High-Z, Generator says "High-Z"

That's the limitation of the generator. You've got the cheapest thing on the market, it will have limitations like this.

Well, that's crap.  I'll just have to program a microcontroller then.   

Thanks for the help

[ataradov]

Into my pimped up (100MHz modded) Rigol 1052E scope with High-Z, Generator says "High-Z"

Well, at 100 MHz bandwith, you are getting only 3-5 harmonics, so it is not going to look square no matter what you do.

Well, that's crap.  I'll just have to program a microcontroller then.   

Something is not right with your setup, it will not go down to 1 uS, but it should be able to do 16 uS according to spec.

[NivagSwerdna]

Well, at 100 MHz bandwith, you are getting only 3-5 harmonics, so it is not going to look square no matter what you do.

Yeap. I'm happy with that one.  I can see the issue there and can live with that one.

Something is not right with your setup, it will not go down to 1 uS, but it should be able to do 16 uS according to spec.

Interesting.  16uS would be fine.

[ataradov]

Interesting.  16uS would be fine.

I'll try this on my device once I get home unless you figure it out by then or someone else jumps in. But even official manual shows default setup at 1kHz frequency, 4.0Vpp amplitude, 0Vdc offset, 200us pulse width.

[ataradov]

It looks like minimum pulse width it depends on the frequency.

At 1 kHz I get allowed pulse width down to 1 uS. Not sure what is actually going to be on the output though.

At 100 Hz, it jumps up to 10 uS.

And at 1 Hz, it is 1 ms, just like you see.

[NivagSwerdna]

Thanks for following up.

Yeap.  Indeed the 1mS minimum pulse width is when the frequency is 1Hz.  I need a slow cycle so I can't dial it up to kHz.

I find it odd to have this limitation.   I'm just asking for a 10uS on time and 999990 uS off time.  Don't understand why that is so hard?

[tggzzz]

Thanks for following up.

Yeap.  Indeed the 1mS minimum pulse width is when the frequency is 1Hz.  I need a slow cycle so I can't dial it up to kHz.

I find it odd to have this limitation.   I'm just asking for a 10uS on time and 999990 uS off time.  Don't understand why that is so hard?

Nitpick: please stop confusing SI units of time (seconds, s) and conductance (Siemens, S). While it is clear here, in some situations it is ambiguous - any you might as well learn good habits sooner rather than later.

Yes, I know samples are sometimes written as "S", but that is not a SI unit!

[ataradov]

Don't understand why that is so hard?

They use DDS with a fixed memory depth. And in order to create longer period, you need to "play back" that buffer at a lower sample rate. To create a pulse you need at least two samples, so if you decrease sample rate, you increase pulse width.

The only functions that are implemented separately are sine and square waves. They can run at full 25 MHz as claimed.

[NivagSwerdna]

Nitpick: please stop confusing SI units of time (seconds, s) and conductance (Siemens, S). While it is clear here, in some situations it is ambiguous - any you might as well learn good habits sooner rather than later.
Yes, I know samples are sometimes written as "S", but that is not a SI unit!

Consider that nit picked.  Lazy capitialisation.  I meant "s"... I really really meant "s".

[NivagSwerdna]

They use DDS with a fixed memory depth. And in order to create longer period, you need to "play back" that buffer at a lower sample rate. To create a pulse you need at least two samples, so if you decrease sample rate, you increase pulse width.

Ah. Thanks.

[karoru]

Nitpick: please stop confusing SI units of time (seconds, s) and conductance (Siemens, S). While it is clear here, in some situations it is ambiguous - any you might as well learn good habits sooner rather than later.

Yes, I know samples are sometimes written as "S", but that is not a SI unit!

It always drives me nuts that old analog scopes tend to display something like "1 uS" on the screen. Yes, I know, siemens got introduced in 1971 and electronics people were always a bit flamboyant with unit symbols, but still (there were even digitizing HP scopes built in the eighties that didn't acknowledge that SI yuppie stuff)...

I wonder what's newest available test gear from big vendor still using "kilomegacycles"

[tggzzz]

Nitpick: please stop confusing SI units of time (seconds, s) and conductance (Siemens, S). While it is clear here, in some situations it is ambiguous - any you might as well learn good habits sooner rather than later.

Yes, I know samples are sometimes written as "S", but that is not a SI unit!

It always drives me nuts that old analog scopes tend to display something like "1 uS" on the screen. Yes, I know, siemens got introduced in 1971 and electronics people were always a bit flamboyant with unit symbols, but still (there were even digitizing HP scopes built in the eighties that didn't acknowledge that SI yuppie stuff)...

I wonder what's newest available test gear from big vendor still using "kilomegacycles"

Tektronix was the big offender with "uS" - then they started producing digitising scopes, and realised that 100S/S wasn't a good marketing spec

And don't forget "uuF" for capacitors. Still seen in the manuals for vintage scope probes, just to trap the unwary fools who are trying to refurbish them. (Please don't ask how I know that, because I'll only get annoyed and have to find a way of avoiding answering the question )

[g0hjq]

(1) I tried to generate a 25MHz Square wave.... it doesn't look at all square!  Is that because I would need a much higher spec to generate such a signal?

Have you tried pushing the scope probe directly into the centre hole i the Signal Generator's BNC socket and shorting the probe earthing band to the BNC outer? It's a bit dirty I know, and might eventually wear the BNC, but I do see a recognisable square wave when I do that on my SGD1020. I wouldn't expect it to be anything like perfect on a 100MHz scope though - you'd need double that to see the shape accurately.

(2) I tried to generate a 1uS pulse with 5V hi, 0V low at a period of 1Hz.  It won't let me do this... it says the minimum pulse width is 1mS!  That's 1000x larger than what I wanted!

I get the same. It seems to be a limitation of the SDG1020 that the ratio cannot be more than 1000:1.