• Check out the Annual AK Secret Santa exchange Program, signup before November 23rd in the AK Charities & Christmas forum!

Newbie square wave question

Set the frequency generator to 1 KHz and adjust amplitude to so that you read 4V peak on the scope across a 8 ohm non-inductive dummy load resistor on the 8 ohm speaker tap. That’s your base line measurement, I.e 0 dB.

Then without changing the amplitude, dial the frequency down. See what frequency it reaches 3.56 V peak amplitude. That will be 1 dB down relative to 1KHz.

Then without changing input amplitude dial the frequency up until again you read 3.56 V peak amplitude. Again that will be 1 db down relative to 1 KHz.

If you’re doing this open loop (no feedback or HF compensation applied) you should see the -1 dB frequency response points at something like 30 Hz to 10 KHz or so.

Then apply say 10 dB feedback and repeat the process. You should see frequency response bandwidth increase to something like 10 Hz to 30 KHz.

For power bandwidrh measurements you need to run the amp at 90% of max power into a large 8 ohm dummy load and repeat the measurements. A bit harder to do since the dummy load will heat up a lot faster so you need to be quick on the measurement and then remove signal and let the dummy load cool.

With 10 dB global feedback you should see power bandwidth 35 Hz to 20 KHz, with Hammond output iron.
 
Last edited:
A square wave is made up of the fundamental frequency and odd harmonics, so in the case of a 10K square wave it is 30, 50 and 70 kHz and above. You need at least a reasonable response to the 30kHz and hopefully the 50kHz to make a decently square wave. Negative feedback can help to improve the look of the signal.
 
The poor rise time in your scope shot indicates a loss of transient response. It's not terrible, and it looks pretty stable, but it's inferior to what you'll get with an original Peerless. They turn up on the Bay pretty often. Another option is the Heyboer/Peerless S-265-Q with 50% screen taps. They perform very well, can be used in the W5 circuit and are quite affordable at about $500 a pair.
 
It is always a gamble on Ebay. Many times you get someone else's problems. That said I buy on Ebay a lot myself.
 
When doing the baseline test with global feedback should I use the 16 ohm tap like the original or should I move it over to the 8 ohm tap ?
 
I understand that the bandwidth is limited in comparison to the original transformer however I still want to extract the best performance. At some point I would love to do comparison and see just how audible the difference is. When I last listened to it the sound was actually pretty decent and not lacking In bass. The upper frequencies seem to be good but it did not have that air. The limitation could have also been my preamp.
 
Are you using the original value feedback resistor size? If yes you’ll want to use the 16 ohm tap. Otherwise (with Hammond classic wind secondaries) you will want to use the 4 ohm tap. Using the 4 ohm tap will require the original value feedback resistor size to be halved if you want to apply the same amount of feedback as the stock circuit.

I would not pull the feedback from the 8 ohm tap when using Hammond classic wind secondaries. Doing so could create stability issues.
 
Expect that the cap value across the feedback will be different too, but 1/2 the resistor generally goes with 2x the cap value. Thats probably going to be wrong for a non-original transformer. If there is a shelf network off the voltage amp stage, those values will probably be wrong too.

If this has a fair size grid stop resistor between the input jack and the input grid, that also produces slower rise times.
 
The feed back and steps have been removed for testing. All the component values are the same as original with no circuit modifications. Do I need to calculate my amplifiers gain to find the correct component values for the feedback circuit ?
 
You can do it that way, but I use a far lazier method. Set up the amp with no feedback, wind it up to say 3 volts of output. Actual value is not important, just something easily measured but well below clipping. Figure out how much feedback you want, and work out what that many db down from your set voltage is. Connect a pot and dial it to that voltage. Unhook the pot and measure the value. Install the closest resistor to that value.
 
Without any other guidance I'd probably use something around 2x what the original value was so there is room to go either way.

I'm interested to see if that logic holds up. :)
 
So adding 10 dB of feedback can make quite a difference to the bandwidth. I'm wondering how much the output transformer affects the final results. Would a different output transformer give noticeably different frequency response measurements?
 
Without any other guidance I'd probably use something around 2x what the original value was so there is room to go either way.

I'm interested to see if that logic holds up. :)
sounds reasonable. No idea what the original resistor value is, so guessing maybe a 10K pot? Its not overly critical, just too large a pot makes for a touchy adjustment, and too small may just be below the value you need. easy enough to tell, if its set to max resistance and drops below where you need it, the pot is too small. Swap for a bigger one, or add a resistor and measure the pot+resistor value.
 
10x probes are not needed to test the output. They are sometimes useful for testing high-impedance signal levels. Typically, the 10 kHz square-wave output is viewed at an equivalent output of 1W, which is 4v into 8 ohm load. Easier to see at 2v/div. Does your scope auto sets for 10x probe? If not, you are viewing 10x the signal level, so a 10 kHz square wave will never look as expected.
4 volts into an 8 ohm load is 2 watts:

P=V^2/R = 4^2 / 8 = 16 / 8 = 2

For 1 watt into 8 ohms you need a voltage of 2.83 volts.

Shelly_D
 
I think this may help this is the schematic I have been following.
 

Attachments

  • IMG_2591.jpg
    IMG_2591.jpg
    176.9 KB · Views: 20
4 volts into an 8 ohm load is 2 watts:

P=V^2/R = 4^2 / 8 = 16 / 8 = 2

For 1 watt into 8 ohms you need a voltage of 2.83 volts.

Shelly_D

Somewhere along the way someone mentioned peak and peak (or peak to peak) is what you see in a scope waveform.

4Vp (or 8Vpp) = 2.83Vrms = 1W @8 ohms.
 
Last edited:
FWIW, I think it`s best to monitor the square wave generator`s output on another trace for comparison (requires a multi-channel scope) while doing these tests to make sure the square wave generator amplitude, and wave form are not shifting and becoming frequency/amplitude problematic, which requires another scope probe that`s been compensated to the other scopes channel.

I learned the hard way in 1978 while wasting time frustratingly trying to track down a apparently non issue with the amplifier I was testing due to the trusted, or so I thought, waveform generator I was then using, started having internal issues with it waveform purity unbeknownst to me, until it was suggested by my boss to scope the waveform generator` output signal !
That monitoring of the waveform of the signal generator in use on one of the channel`s of my dual trace scope(s) became standard bench practice from that moment on, especially important with the use of square wave signals above and below 1 kHz, as some less quality wave form generators can impart their own artifacts to the square wave form at both the selected fundamental higher & lower frequencies.

Trust, but verify.

Just something to take in consideration.
 
Thanks I’m using both Channels the slope of the waveform appears to increase as I approach 6k when using a square wave. I was hoping it was something minor that I overlooked.
 


Thanks I’m using both Channels the slope of the waveform appears to increase as I approach 6k when using a square wave. I was hoping it was something minor that I overlooked.
Yeah, it will, the higher the frequency.

But, I thought you were going to map the frequency response using sine waves and amplitude?
 
Back
Top Bottom