A note about your AF Generator 01 (retro design)

Hi Ruud,
That sounds interesting. It is exactly the same method as the "SINADDER" instrument I used in my days in the PYE service dept.


Sinadder-2



Sinadder-3

I would have thought that to get a notch better than 50dB would be a tall order. The sinadder-2 was only good to about -35dB or so. The Sinadder-3 was better. I somehow trust the spectrum analyser method as it is easy to see if there is any rubbish there, whether it be caused by the instrument or the signal being tested.

If I was "into audio" then I would probably spend a bit of time to make a decent instrument.

But thanks for the information. The Nakamichi sounds like a nice instrument.

Best regards from Harry - SM0VPO

Ruud wrote:You mention that you get about 0.2% distortion. That is a very good value in most cases.
By the way: how do you measure distortion?
For years I have used my good old Nakamichi T-100 audio analyzer, to measure distortion.
The only disadvantage is that the T-100 can only measure distortion at 400 Hz. (0.01% is the lowest value that can be measured.)

Hi Ruud,
I am very limited on test gear, so I rely a lot on apps for the mobile telephone. That is why I am back-to-basics building my own equipment.

I thought about distortion and noise measurements and decided that noise is too wide-band to get a meaningful reading if it is sufficiently low. But distortion is a lot easier. I used an AF spectrum analyser, but I also had to use a low frequency tone to get a meaningful reading. Distortion creates harmonics and I figured that measuring the harmonics was the easiest.

I tried about four different apps on five different Android telephones to compare the differences. otherwise you could just be measuring the telephone's input circuitry. Huawei had a high noise floor but both my Galaxy phones gave the best results. -53dB is about the limit of measurement, which i thought was excellent. Perhaps it could be better? But it is a good indication, anyway. It is in the right area.

I have never been into audio quality, other than meauring SINAD in radio receivers, or complaining about crossover distortion in my Bush Arena HiFi amplifier left channel. But that was a few years ago.

BR Harry - (SM0VPO, still trapped in Spain)

Hi Ruud,
I evaluated a SSB demodulator, where low distortion was crucial, at the beginning of my career. I used a distortion meter from Poland, self-tuning across the audio band. I do not remember its type after all those years. I expect using an FFT analyzer (soundcard and software) is the best solution nowadays.

VBR from Ivan

You mention that you get about 0.2% distortion. That is a very good value in most cases.
By the way: how do you measure distortion?
For years I have used my good old Nakamichi T-100 audio analyzer, to measure distortion.
The only disadvantage is that the T-100 can only measure distortion at 400 Hz. (0.01% is the lowest value that can be measured.)
Because I have been designing a lot of audio equipment for audio "phreakz" (they are the worst!), at a certain point I felt the need to measure distortion at other frequencies as well. I knew that there was a Hameg distortion meter, that could measure from 20 Hz to 20 KHz, but that HM8027 was out of production. After many years of searching, I found a used one in good condition on eBay.
But the next problem became soon obvious: if you want to measure very low distortion figures, you will need a low distortion generator.
Fortunately, the Hameg HM8037 was still available at the time. This generator can produce a sine wave with max. 0.01% distortion.
Unfortunately to operate this Hameg combination, you needed a 'base frame', with nothing else than a power supply in it. 
A pretty expensive thing, if you ask me, but it has been worth the money. 
I can now measure distortion values down to 0.01% over the whole audio frequency range.
And isn't it so, that from time to time you have to give yourself a little present?...

All the best from the Netherlands, and stay away from the corona virus!

Admin wrote:... the total current consumption for the oscillator is about 14-18mA. That is really interesting, as it means the current through the 6V, 40mA lamp is only about 15mA or so. ...

Hi Harry,
tungsten filament lamps are said to have the best regulation properties when their glow can be barely seen in the darkness. So 15 mA for a 4O mA rated lamp is probably optimal.

I wish you better luck with your travelling. I am still disabled with my backbone, so I do not feel all those coronavirus limitations (wearing a respirator in public, limited transportation etc.) as too hard.

VBR from Ivan

Hi again Ruud,

I built the generator before I came out here, but now there is nothing else to do during the coronovirus quarantine I did some experiments and measurements with the generator.

It seems that to get 1.5V Pk-Pk out, the total current consumption for the oscillator is about 14-18mA. That is really interesting, as it means the current through the 6V, 40mA lamp is only about 15mA or so. That means I don't need all those buffer transistors.

I have been tidying up the PCB to reflect the changes. It looks really simple now. just as it originally did with the R53 10mA thermistor last century  
With fewer components it should be cleaner, but the NFB would have kept distortion down a lot.

Today I experimented with "silk-screen" layers on printed circuit boards, using a laser printer. When I get back to Sweden I will be making a much better quality PCB

Yesterday we were told that Norwegian airlines had cancelled our flight home next week, and we cannot get another flight for another month. The flight is still scheduled. I suspect that they just kicked the ecconomy travellers off and put full-fare passengers on. I was a bit depressed. Registered with the Swedish Embassy and consulate that were were stranded. Today I booked the last two "business class" (full-fare, luxury, including food) tickets with a different airline. We have to spend a night in the airport, but we will be home only a day late. You can solve almost any problem if you throw money at it (£1000)

But still cheaper than paying for a private local doctor for chemotherapy meds, local rental-car for a month, black-market toilet rolls, etc.

Very best regards from Harry - SM0VPO (In Spain)

(PS - just had an Avast popup notification about corona virus )

Hi Ruud,
Just a quick note to tell you that this evening I made up the PCB for this project. I put 10Ω resistors in the emitter of each transistor.

I was also a little concerned that the voltage drop across the biasing diodes could cause the push-pull pairs to conduct a bit too much standing current, but it also followed your suggestion to limit the output current. The finished project only takes about 12mA, and most of that is the current through the lamp.

The PCB on www.sm0vpo.com does have a couple of errors, but they have been corrected. Just need to upload new pictures. I did a bit of hacking on the board and it works fantastic  

I dropped the 100Ω in series with the lamp to 68Ω and it now delivers 4V pk-pk. With 100Ω it started clipping at 11.5V supply. Distortion is better than 0.2% with supply voltages from 7.5V upwards. It also has enough output to drive a small 75mW Christmas tree lamp (0.07W).

So now the prototype is done I scan re-build it without the updates, then box it and update the homepages.

My new home-made PCB drill, light-box and PCB-CAD program have all been put to use today. Not had these facilities for a long time, so I am really happy today. Thank you for your feedback.

Best regards from Harry

Ruud wrote:I think the whole point of the design is to have a low distortion sinewave.
(So no square wave.)

Hi Ruud,

Yes, the object of this design is a low-distortion sinusoidal waveform. But I hope that I can use it for a vaiety of purposes.
I will see if I can have this project modified and assembled in the next two weeks.

Another home "open day" next weekend, so the living and cleaning continues!

Best rgeards from Harry - SM0VPO

I think the whole point of the design is to have a low distortion sinewave.
(So no square wave.)

hi,
it happened that i wanted to test a light organ this days. Since i had no AF generator i built one with an arduino... Its not a big deal but i can draw and share it if someone thinks so.
It goes from 1hz to 65536 hz. Square wave of course. The output comes from the collector of a tranzistor which is tied to 15v through a 1k resistor.
Frequency is set from a pot and displayed on an lcd.
It did the job for me but the arduino and the display are quite under used so other features can be added if needed. Like f meter, electronic voltmeter... or who knows.
First i wanted to reuse the parts but i throw it in a box last day because it was useful and i didn't have an oscillator anyway.

Hi Ruud,
"Just a thought" Thats a very good thought. To be honest I had not thought about it.

When I built the original I used an OpAmp to drive the output, but they have a 20mA current limit. As you state, it could be a transistors destroyer if the hfe and bias is sufficient to deliver a destructive current.

I will take your advice and modify the circuit a little. But it could be good to have at least 100mA output current availability - sufficient to test a loudspeaker, speaker, drive a Cockroft ladder, or annoy the neighbours ;-)

Best regards and thanks for the feedback. Really useful for a 69-year old brain :-)

Hi Harry,

I took a look at your AF retro generator.
This type of oscillator can produce a very clean sinewave. I have used this principle before.
But...I have a little problem with the 2-transistor output stage without any form of protection.
Personally, I use the principle: OpAmp, two diodes, NPN/PNP output stage for a couple of headphone amplifiers in my studio.
As you mentioned before, it is possible that the current through the output transistors will be higher than 40 mA.
With a bit of bad luck, this could be the ideal transistor destructor!
To prevent this, I always use a 33 ohm resistor in each emitter leg.
This will prevent high current through the output transistors.
When the current through the transistor becomes higher, the voltage over the 33 ohm resistor will become higher too, and the Vbe of the transistor is automatically reduced. If the current would be 200 mA, the voltage over the 33 ohm resistor would (theoretically) be 0.66 V and this would prevent the transistor(s) from conducting. With the 33 ohm resistors I never had  'self destructors'!
And it doesn't really degrade the performance of the circuit.
Well, just a thought...