Hi All,
I've just obtained one of these old distortion analysers and it works incredibly well for something built in the 70s and quite probably unserviced since then.
I've managed to get some sensible measurements from it and am following the "Greening of an IG-18" along with it quite nicely.
It has an output which i am plugging into scope at the moment for some basic FFT but would like to pipe it into a Tek 2712 spectrum analyser I have for high frequency work.
The issue is that the output would like to see a 600ohm load and the spectrum analyser has a 50 ohm input resulting a collapse of the voltage level of the signal from the HP to a point where it looks very noisy and ugly on the scope.
Being a noob at all of this I'm not sure what i need to correct the mismatch, I'm guessing some sort of active buffer amplifier and am thinking this might be a good place to put some of the filters that the HP lacks, chiefly a LP filter.
I'd like to have a stab at a DIY product if at all possible, but if something cheap, easy and high performance is out there I would consider it too.
I've just obtained one of these old distortion analysers and it works incredibly well for something built in the 70s and quite probably unserviced since then.
I've managed to get some sensible measurements from it and am following the "Greening of an IG-18" along with it quite nicely.
It has an output which i am plugging into scope at the moment for some basic FFT but would like to pipe it into a Tek 2712 spectrum analyser I have for high frequency work.
The issue is that the output would like to see a 600ohm load and the spectrum analyser has a 50 ohm input resulting a collapse of the voltage level of the signal from the HP to a point where it looks very noisy and ugly on the scope.
Being a noob at all of this I'm not sure what i need to correct the mismatch, I'm guessing some sort of active buffer amplifier and am thinking this might be a good place to put some of the filters that the HP lacks, chiefly a LP filter.
I'd like to have a stab at a DIY product if at all possible, but if something cheap, easy and high performance is out there I would consider it too.
By "DIY product," do you mean one of your own creation at the component level?
If I were starting from scratch, I'd use a device I designed with just a couple years ago: OnSemi SMA3107. They're cheap, 50 ohm in and out and have flat gain (~ 20dB) to DC, limited only by coupling caps. Modest noise figure that will be good enough for this task. Probably a cascade of two devices, but only one might suffice. You could insert a ~ 560 ohm resistor between monitor output and gain block input to "impedance match" if you care and have sufficient gain. The acid test for adequate gain is to note spectrum analyzer noise floor at max sensitivity, then apply input from THD analyzer. Noise floor should rise so the noise floor is dominated by noise from the THD set.
Other members may be able to suggest an Ebay solution.
If I were starting from scratch, I'd use a device I designed with just a couple years ago: OnSemi SMA3107. They're cheap, 50 ohm in and out and have flat gain (~ 20dB) to DC, limited only by coupling caps. Modest noise figure that will be good enough for this task. Probably a cascade of two devices, but only one might suffice. You could insert a ~ 560 ohm resistor between monitor output and gain block input to "impedance match" if you care and have sufficient gain. The acid test for adequate gain is to note spectrum analyzer noise floor at max sensitivity, then apply input from THD analyzer. Noise floor should rise so the noise floor is dominated by noise from the THD set.
Other members may be able to suggest an Ebay solution.
Hi Steve, thanks for chiming in. I’m not up to designing a solution myself without help. I do want to learn and become more capable so I’ll look that part’s data sheet up and lay something out on diptrace. Would be, as ever, grateful for your help with it.
I had assumed there would be some turnkey solution to this kind of thing but would enjoy building my own for certain.
Just looked, I don’t need any gain, I think.
The residual from the analyser is 100mV max, the spectrum analyser doesn’t like more than a volt or two and so a gain of more than 10x would be too much, wouldn’t it?
I had assumed there would be some turnkey solution to this kind of thing but would enjoy building my own for certain.
Just looked, I don’t need any gain, I think.
The residual from the analyser is 100mV max, the spectrum analyser doesn’t like more than a volt or two and so a gain of more than 10x would be too much, wouldn’t it?
Last edited:
Maybe you already have enough gain? Try the experiment I mentioned above. It's to be expected that the displayed SA result would be noisy and ugly. After all, you're using the SA to dig into the noise floor of the THD set where all the imperfections are hidden in its analysis bandwidth of a few hundred kHz. SA resolution bandwidth can be as low as a few kHz, allowing the hidden junk to emerge from the noise.
What is the low frequency limit of the Tek? If you ate looking at the residual harmonics at the output of the 333 then they will mostly be in the audio band.
The 333 output really needs a high impedance input like more than 10K Ohms. A high Z input headphone amp may be a good simple fit for this.
The 333 output really needs a high impedance input like more than 10K Ohms. A high Z input headphone amp may be a good simple fit for this.
I checked and the lower limit of the 2712 is 9 KHz. That means the lowest fundamental you would look at is 4.5 KHz. A soundcard plus REW or ARTA may be more useful. It does not need to be exceptional since the HP removes the fundamental so you don't need a lot of dynamic range.
Thanks Steve, perhaps a better way of putting it would be that the output from the HP is nice and clean into a High Z oscilloscope input, but the same signal looks very much degraded when fed into the SA and monitored with the oscilloscope.
I have a Levell micro volt meter that has a low noise preamp, I’m sure that could drive a 600 ohm load at least. I will try with the resistor and see what sort of signal I get out of it.
Thanks buddy, this is the plan with the lower frequency stuff but I also fancy having a poke around the 20Khz thd and imd specs of a few of my amplifiers and for that the SA seems ideal.
I like the idea about the headphone amplifer, seems ideally suited to driving 50 ohms or under.
I could even build my own I suppose, a little class A thing that might constitute my first scratch built amplifier in fact... 🙂
P.S.
Does this thing need re-capping? It takes a good hour to warm up and a couple of the voltages were off by as much as 30% before I adjusted them to service manual specs.
Does this thing need re-capping? It takes a good hour to warm up and a couple of the voltages were off by as much as 30% before I adjusted them to service manual specs.
Of course, that High Z scope input is a very light load, very wideband, and not very sensitive.
I couple more experiments suggest themselves.
I'm assuming the 333A monitor output has 600 ohm output Z. If you monitor that output with your scope and then shunt the probe with a ~50 ohm resistor, anything unexpected other than the big 12:1 reduction in amplitude? There are added equipment grounds being introduced...
And a similar experiment: if you inspect the SA input with the High Z scope, any surprises on the scope or on the SA screen?
And one more: If you examine the SA screen before and after attaching the scope at the SA input, any changes on SA display?
Simple solution is 560r in series with input. Level will drop of course.
A little more complicated solution is low impedance output op amp buffer. No level drop.
A little more complicated solution is low impedance output op amp buffer. No level drop.
I just checked. The output has the meter circuit in series with it. You will probably get lots of harmonics from it. You may be able to add a buffer to get the distortion out upstream of the meter circuit.
Recapping? not sure, I remember the 333 can be groggy after a long hibernation. Thought it was the switches. 30% in power supplies variation is large. Try leaving it on a few hours a day and see if the caps get excited. The 339A unit, the noticeable difference is faster in auto tuning the test frequency as you lower the meter distortion scale.
The 9kHz lower limit doesn't mean it doesn't work down to DC, just that the response rolls off due to the input DC blocking capacitor (1uF typically). You'll see -0.5dB at 9kHz, -3dB point at about 3kHz.
BTW that 1uF cap is ceramic and non-linear - not the remotest issue for RF as no signal voltage appears across it, just the DC offset.