in the search for a High precision SIG GEN

[sdouble]

Dear all,
I'm currently looking for a high precision signal generation.
I'm only after square waves.
Rise time and decay are important : 2ns 10-90% at worst.
The max frequency is not an issue : I typically operate from 1kHz to 1 MHz.
What is really important is the amplitude accuracy.
I need to explore the response of my circuit to amplitude varying from 5 mV to 5 V.
The major issue is to reach the needed accuracy in the 5 mV to 500 mV range where 0.5% accuracy is needed
Any clue about the device which could fit the bill ?
I need such a device for a work project. I'm not looking for a cheapo.
Thx for your help

[sdouble]

no idea ? Anybody ?
My specs are too demanding ?

[nctnico]

What you are after is a network analyser but an oscilloscope with frequency analysis (FRA / Bode plot) will also do. In both cases the signal going in is compared against the signal going out. So the absolute level doesn't matter. Only thing you'll need to check is gain error between channels which is easely performed by connecting both input channels to the output and see if the measure amplitudes match within your required specification (if not, you could even compensate for the error).

[egonotto]

......
The major issue is to reach the needed accuracy in the 5 mV to 500 mV range where 0.5% accuracy is needed
....

Hello,

that will be a problem. 0.5 % of 5 mV is 25 uV. Noise already plays a role here.
Even the almost 10000 € expensive Siglent SDG7032A only has:
"Amplitude accuracy ± (1% + 1 mVpp) (Condition & Note: 10 kHz sine, 0 V offset)"

Best regards
egonotto

[noisyee]

For your reference, the oscilloscope calibrator golden standard fluke 9500b, it's square wave amplitude accuracy is 0.1%, but only from 10 Hz to 10 kHz.
0.5% amplitude accuracy for 1MHz square wave is a demanding requirement. That means ~0.5dB flatness of accuracy up to ~10MHz. Some function gen may have that level of response flatness but not always been specified in its datasheet.
Are you overlooking the accuracy requirement? Or you really need is a nearly flat square wave? If absolute accuracy is not that important, I'd recommend Keysight 33500B or Siglent SDG6000X, which I had verified its square wave flatness.

[Phil1977]

As soon as you require information about high precision products, it´s practically impossible to give advice without knowing the application.

If you just want a 12V 5A power supply and tell me you want the cheapest or the highest efficiency, then I could give you some advice without knowing if it´s for powering your laptop or charging a battery. But if you want something with the accuracy you mention, it´s absolutely necessary to consider the whole system the signal generator is embedded in. So please state some more application information! (and hopefully it´s not about body chakras again...)

Best, Phil

[tautech]

Siglent SDG6000X, which I had verified its square wave flatness.

Just meets the 2ns requirement......

[sdouble]

Thw all,
I do realize, that my post was kind of misleading.
I need a well calibrated square pulse generator to inject a charge sensitive amplifier through a capacitor.
Amplitude x Capacitance gives the injected charge.
I need to ram up the injected charge from 5 mV to a couple of volts to check the spectroscopic chain linearity.
If the injection signal (the square pulse) is not linear, I'm screwed.
The ripple/noise is not really an issue here since the charge preamp output will go to a spectroscopic amplifier, which is basically a band pass filter.  (300Khz typical central frequency) and a ADC to get the peak amplitude.
I'll look at the average amplitude over , says thousands of triggers, washing out further the uncertainty linked to rippls/noise.

[jonpaul]

Bonjour cher  Monsieur, si vous habite a La France nous pouvons discuter tes probleme interessant.

SVP m'informe tes contacts avec PM

Jon

Hello dear Sir:  if you live in France we can discuss your interesting problem.

Please inform me of your contacts with PM

Jon

[noisyee]

I need a well calibrated square pulse generator to inject a charge sensitive amplifier through a capacitor.
Amplitude x Capacitance gives the injected charge.
I need to ram up the injected charge from 5 mV to a couple of volts to check the spectroscopic chain linearity.
If the injection signal (the square pulse) is not linear, I'm screwed.

If I understand correctly, you don't need absolute amplitude accuracy, you need a known accuracy amplitude reading.
Maybe you should consider using a higher grade instrument to calibrate reading of your not so accuracy square wave gen. A quality DMM like Tek DMM7510 will guarantee you good accuracy up to 100kHz. PicoScope 4262 specific ±0.25% DC accuracy, although FR flatness not specific, it should have decent performance up to some MHz. A decent ADC EVAL board or even Digilent Analog Discovery should have that level of linearity you need if you don't care about accuracy.

[sdouble]

that's a very good idea.
A good quality Multi in AC Mode should provide me with what I'm after, right ?