HP 33120A output impedance

[BMK]

I'm attempting to use a HP33120a AWG as a calibration source to verify frequency response of an ADC in a machine vibration measurement device. The idea is that the calibrated AWG is to be used in place of a physical vibration standard (shaker table) to quickly verify the measurement electronics without dismantling the transducer (iepe, piezo with inbuilt buffer amp). Anyway I just noticed that the amplitude setting on the HP assumes a 50R load impedance and appears to be open loop, meaning I can't  believe its displayed amplitude when terminating with resistance other than 50R. On HighZ mode, my ADC loads the signal noticeably. We are dealing with frequencies only up to 5kHz over a couple of metres RG56 so I'm ignoring transmission line effects. A 100Hz 1v RMS sine wave gives significantly different (+/- 5%) when measured simultaneously with a Rigol 1054z and a Fluke 87iv.  So I have 3 numbers that don't match despite all instruments being in Cal.
ADC input impedance is 100k.
Now questions.
-For the sig gen driving my ADC, am I better to terminate AWG with 50R, or to just take double the indicated amplitude as the real value?
-If I terminate, I assume I need a precise 50R terminator, what's the common practice?
-Fluke and Rigol disagree by 5%, any ideas what is going on, perhaps I'm expecting too much of the TRMS calculations in the scope?

[TheSteve]

I would terminate with 50 ohms - at lower frequencies I would think a 50 ohm 1% resistor will do just fine.

[TiN]

It's 50 ohm output.

[bson]

Just terminate with a 50ohm load.  50 || 100k = ~50

Or you could insert a simple follower for vanishingly low source impedance as seen by the ADC.

Or measure the input voltage and use that value instead, or dial up the 33120A until you get the desired amplitude.

[baoshi]

I compared the RMS reading on my DS1052e vs HP34401A, connecting to the same source 33120A. The difference is about 1v vs 1.04v.
Consider the recommended calibration equipment for 33120A is 34401A, I would happily trust 34401A's measurement.
I wonder if Rigol DSO has a true RMS converter or it is calculated by accumulating. A nice finding is done by Kerry Wong at http://www.kerrywong.com/2011/08/22/on-rms-voltage-measurements/

[sorenkir]

Output impedance is 50 ohms, but you can set if you are terminating the output or not in SYS MENU > OUT TERM (User's Guide p.40)
Michel.

[BMK]

Thanks for all the replies,
Rtfm indeed. In High z mode the AWG is still has output impedance 50R.  I was wrongly interpreting the meaning of HighZ mode, thinking that it actually altered the output impedance to some Kilohms- duh. I thought the strange RMS values measured were because my 100k ADC was loading the AWG. Turns out Fluke 87iv is pretty much bang on with the AWG. But I can replicate baoshi's result so I guess the only puzzle is why the Rigol 1054z is so far off on RMS for a 100Hz sine wave.