Tweaking resistors in SMR4 resonance compensation

Hi!

I want to boost the resonance compensation on the Ambika SMR-4 filter so that a larger portion of the original signal is feed back into the filter when resonance is increased. I’m thinking that the two resistors R29 (10kohm) and R30 (220ohm) function as a voltage divider and I would achieve the boost by increasing the ratio R30/R29 (e.g. by a factor of ~2-4).

Is there any practically relevant difference between lowering R29 or increasing R30 in this circuit? Does the difference in power over the resistors matter?

I’m thankful for any advice!
B

You’re looking at the wrong components. Start by increasing C22 to 4.7µF (you can use a bipolar electrolytic for this task). With the value of 220nF, the compensation gets weaker as we go lower and lower from about 300 Hz – something I did on purpose to avoid overwhelming bass.

Once you fix that, the resistor ratio is already well-chosen to keep the pass-band gain more or less constant when resonance is increased. Using a different ratio will create a very silly effect where the overall signal amplitude is increased as you turn resonance up.

2 Likes

Thanks Émilie!

I haven’t tried to increase C22, but I did try to bypass it completely (after reading up on SMR4 development and SDE) and that didn’t boost the bass in the resonance compensation at all to my ears.
That’s why I started thinking about boosting the level of the original signal in the resonance compensation feedback.

Edit: failed attempt of resonance compensation bass boost here: Another modified Ambika SMR-4 voice card

Cheers
B

Have you tried with a larger cap?. Not having a cap at all could cause DC offset issues, which might be… counterproductive.

Have check on a scope by how much the signal in the low frequencies is attenuated when resonance is increased?

I don’t have a scope but when I have time for empirically tests I’ll try to use a bigger cap in my modified SMR4 and do some recordings and a proper analysis.

Thanks for the advice!

Ok,

so I tried out two things to get a little beefier sound with resonance:

  1. raising C22 from 220nF to 4.7uF to reduce the amount of high pass filter in the resonance feedback compensation
  2. raising R30 from 220 ohm to 1 kohm to increase the overall level of resonance compensation

Based on a simple listening excercise with varying filter cutoffs and resonance I conclude that option 1 doesn’t really change much, and option 2 is more to my liking, i.e. the overall bass is maintained with increasing resonance. The increase from 220 to 1000 ohm is a bit excessive, but I don’t find the overall level changes silly. Attached are sound snippets of the two modifications compared to the original.

So now I still wonder if it makes any practical difference to increase R30 or decrease R29…

Cheerssmr4mods.zip (1.5 MB)

1 Like

Hi again,

I’m not sure if anyone’s interested, but here are some spectrum plots comparing different resonance levels for the original resonance compensation (R30=220ohm) and the boosted resonance compensation (R30=1kohm). Filter cutoff was fixed at 96 and the resonance was varied from none, to mid, to max.

It’s quite clear that the low range spectrum drops significantly when raising the resonance with the original compensation, whereas it is steadily maintained with the boosted compensation. Of course, with the resonance peak added, the overall signal level is slightly increased with the boost, but not much.

My take on all of this is that it depends on what you expect when increasing resonance for a low pass filter. I guess my subjective expectation for a low pass filter is that if the filter cutoff is higher than the low end frequencies I expect their level to be reasonably steady if I increase the resonance. This allows for adding that nice fizziness without loosing the low end.

Cheers






1 Like