R&S SML01 1.1Ghz Generator

[G0HZU]

Like I said, a lot depends on what you consider to be acceptable. I think about  +/- 0.7dB levelling is typically possible (at the end of a short RF cable) with a fairly modern mid range 1GHz sig gen. So if this is acceptable then this should be fine. If you have a decent RF attenuator you can improve the source match of the sig gen but you would ideally have to know the attenuator was healthy in terms of VSWR and flatness to 1GHz.

For a bit of fun, I powered up an old Marconi 2019A sig gen. This operates up to just over 1GHz. This sig gen cost me about £120 over 25 years ago and I rarely use it. However, it does support external AM modulation with DC coupling.
I also powered up an old Anritsu power meter I bought in the early days of ebay. This cost £200 complete with a thermocouple sensor.
I also used my homebrew levelling amp to level the output of the 2019A sig gen via the external AM mod input. The levelling amp is just a cheap opamp in a box with a potentiometer that sets the output level. This was made from scrap parts and effectively cost me nothing.

The sensor cal shows about +/- 0.05dB for the correction factor all the way to 16GHz.

I also used my homemade version of the 11667 splitter. The resistors cost less than £0.50 to make this. It also needs three RF connectors but these can be purchased for about £5 each.

I then set the output level target to -12.00dBm using the levelling amp. I could have set it to anywhere from about +10dBm down to about -20dBm.

The levelling amp was able to level from a few MHz through to the 1GHz limit of the sig gen. The power meter displayed -12.00dBm at the output of the splitter on every test frequency. But then it should do this because it is in an ALC loop. In reality, the accuracy will be limited by the quality of the splitter and the accuracy of the thermocouple power meter even though the display shows -12.00dBm at every frequency.

The total budget for the above is less than £400. If I had used a 4GHz sig gen then it could have been used to measure 4GHz scopes with low overall uncertainty.

The risk with doing this with 'just' a sig gen is that unless you have a way of verifying that the output is within specification you have to hope that the sig gen levelling is OK.

The Marconi 2019 has fairly good levelling up to 1GHz but I'm unsure of the source VSWR. An external attenuator can improve this, but again, you ideally have to have something to try and prove the level is flat. If this means buying a power meter then you might as well use the same levelling system as me.

A decent homebrew diode detector could just about work to 1GHz if you wanted to (quick and cheaply) prove the sig gen was flat to 1GHz +/- 1dB but this would require a sig gen with low harmonics.

[G0HZU]

The plot below shows the 'typical' flatness of the Agilent ESGD4000A (AKA E4433A) sig gen and this plot is shown in the 'typical' section of the datasheet. The plot shows multiple traces and this shows the flatness of several sig gens on the same plot.

This is very impressive. I've got one of these sig gens here and these E443xA sig gens tend to sell much cheaper than the later E4433B version which looks almost identical.

Note that this plot will be taken using a very expensive Agilent/Keysight power sensor head that will have ultra low VSWR. This effectively removes any mismatch uncertainty due to the source VSWR of the sig gen. A typical scope input will not have a low VSWR up at UHF so some extra uncertainty will creep in when using this sig gen into test devices (like a 50R scope) that don't have an ultra low VSWR across the frequency range of the test device.

My E4433A has the electronic attenuator fitted and there is also a version of these ESGD sig gens with a mechanical attenuator. I've also got the E4433B version here and it has the mechanical attenuator.

Over time the levelling performance will degrade as the sig gen ages but I think it should remain within about +/- 0.2dB. The last time I formally checked my ESGD (a few years ago) it was within +/- 0.1dB on my power meter. The harmonic levels from these sig gens are usually quite low. Harmonics are usually better than -40dBc. I think the spec is -30dBc but it's usually much better than this.

The ESGA analogue versions of these sig gens are even cheaper as they don't have vector modulation. I assume that the flatness performance is similar to the D versions.

It's easy to get in a muddle with the model numbers. ESGD series are the ones with vector modulation. The ESGA don't have vector modulation.

The E4433A is actually an ESGD4000A and this does have vector modulation. I'm not certain but I think all the E443x models are ESGD vector sig gens and all the E442x ESGA series are analogue with no vector modulation.

Prices tend to vary a lot for these sig gens so patience is the key. There are other makes/models with similar performance but they may cost a bit more.

[Martin72]

Hi,

What about this one here:

R&S SML01

[Performa01]

Hi,

What about this one here:

R&S SML01

A strong yes!

Amplitude accuracy of typical 0.3 dB is plenty for all the usual tasks. Phase noise is good and even excellent in "extended divider mode". For characteriszing receivers and spectrum analyzers it's a big bonus to have a low phase noise source like this, even when it's limited to a few dozens of MHz.
It does have sweep mode with up to 1 s per frequency step, hence perfectly usable for letting the scope plot its own frequency response by means of the FFT in peak hold mode.

Only downside is the limited frequency range of just 1.1 GHz. That means, an SDS2504X is about the fastest scope one can effectively characterize with this.

[djsb]

G0HZU,
Have you got any details of the levelling amp you mentioned (schematic)? Thanks.


David.

[alm]

Note that this plot will be taken using a very expensive Agilent/Keysight power sensor head that will have ultra low VSWR. This effectively removes any mismatch uncertainty due to the source VSWR of the sig gen. A typical scope input will not have a low VSWR up at UHF so some extra uncertainty will creep in when using this sig gen into test devices (like a 50R scope) that don't have an ultra low VSWR across the frequency range of the test device.

A common solution is to crank the output of the generator to the highest possible level, and then add as much attenuation as possible at the scope input to mitigate this mismatch.

[Martin72]

Only downside is the limited frequency range of just 1.1 GHz. That means, an SDS2504X is about the fastest scope one can effectively characterize with this.

Short question:
Why ?

[Performa01]

Only downside is the limited frequency range of just 1.1 GHz. That means, an SDS2504X is about the fastest scope one can effectively characterize with this.

Short question:
Why ?

Short answer: whenever I want to characterize a DUT with lowpass characteristic, then I want to look at least one octave beyond the corner frequency, so that I can see the transition into the stop band.

If you use a 1 GHz generator to characterize a 1 GHz scope, then its quite obvious that you can confirm that is within spec up to 1 GHz, but you will never know its actual bandwidth.

[Martin72]

Hi,

Update:
I´ve bought the R&S SML01, should arrive soon..
I'm quite curious about the device, I've never had anything like it.
My siglent HD and I are already "looking forward" to the first tests.
Will continue this thread with first impressions and more, accordingly the title is now changed.

Martin

[Martin72]

Arrived...
At the weekend I have to look in there....

[Martin72]

Hi,

Which connector type is the RF-output ?

Thought it is PL259 and ordered an adapter to BNC - But it´s too big.

[G0HZU]

It's a female N connector (50 ohm)

https://en.wikipedia.org/wiki/N_connector

[Martin72]

Thankyou, now I´ve ordered the right one..

[Performa01]

Thought it is PL259 and ordered an adapter to BNC - But it´s too big.

You will never find PL259 in any serious RF gear that covers frequencies above about 50 MHz. This standard isn't suitable for high frequencies, one reason being that it has an ill-defined characteristic impedance.

RF signal generators traditionally have N-connectors.

[Martin72]

Yes, now I know that too.

[Martin72]

Test it very short - But it works...
It is also interesting how the scope can still "bravely" measure the frequency, up to 800Mhz it is no problem, at 1Ghz it no longer works properly but that is also completely OK.

Next thing is to open it up, I want to see what´s inside...

[Performa01]

... at 1Ghz it no longer works properly ...

Sure. You cannot expect any DSO with full digital trigger system to trigger (hence also display the correct trigger frequency) at the Nyquist frequency and above

[Martin72]

Therefore the " its OK "...
And you can nicely see that the "protection for seeing artefacts" of the scope works when adding the second channel of one ADC.
Having such a "high frequency" source is a good thing to understand how some things work.

[Martin72]

Hi,
opened it up, not much too see there...Superb building quality.
I wonder how that was done with the bandwidth, the 1.1, 2.2 and 3.3 Ghz....
If I should ever get really bored, I'll unscrew more.
For now, I prefer to leave it as it is.

[Martin72]

Rest of the pics..

[mwb1100]

If you still have the MSO5000 could you run some similar quick test whenever convenient to see where and how badly it falls off?

[Martin72]

Hi,

Unfortunately, this is the only model from my "ownership list" that I no longer have access to.
I have access to the following Scope models:

-Siglent SDS2504X HD (Home)
-Siglent SDS1104X-E (Home)
-Siglent SDS2104X Plus (loan from work)
-Rigol DS1054Z (loan from work)
-Rigol DHO804 (Home, soon.. )
-Lecroy WS3024Z (Loan from work)

[KeBeNe]

Hi,
opened it up, not much too see there...Superb building quality.
I wonder how that was done with the bandwidth, the 1.1, 2.2 and 3.3 Ghz....
If I should ever get really bored, I'll unscrew more.
For now, I prefer to leave it as it is.

frequency extension is done by module, which is placed near the attenuator

[Martin72]

Hi,

Want to make some cables....Question:
You can get BNC/N Type connectors for crimping or soldering - Last would be good because I don´t have a crimping tool.
But are there any advantages/disadvantages between the two types ?

[KungFuJosh]

Crimping tools are cheap, and usually returnable if you only need them once. 😉