Sean McLennan's B Blender Questions

[lars-musik]

I am still working on a blending circuit that works with an active high output bass. I had little luck with the Moosapotamus Paralooper. For unknown reasons it introduced lots of noise and after hours of trial-and-error-debugging I am ready to try something new.

I found Sean McLennans BBlender that looks very promising. Now I would like to have a padding down potentiometer at the input to dispose of distortions and a gain pot at the output to make up for the volume loss. I am planning to power the blend circuit with a 7660S charge pump again to minimize distortions in the blender.

Here is the original schematic (had to utilize the wayback machine because the original site is offline so I thought a re-post would be in order):



And here's what I hope works within the parameters given above:




Is the input pad (VR4 500K to 1M Ok?) good and will the Gain pot (VR3 10K-100K) work as I hope?
Thanks a lot for your input!
Yours, Lars

[lars-musik]

So I built this B Blender with an integrated ICL7660S charge pump but still it is quite noisy. The Blender introduces a whole lot of white noise. I tried to replace the TL072 op amps with supposedly less noisy ones (NE5532 and OPA2134 respectively) but then the volume boost (VR3 with a 100K poti) wouldn't work anymore. Just one spot on the pot's travel allowed sound to pass (why is that? With the TL072 it worked fine). Anyway, I hadn't the feeling the noise lessened much.
So – what can I do? Would be some low-pass filtering be advised? If so, where would I place it? Before the "Send" pad? After the "Return"? Or before the gain stage? Or will it all be just fine once I boxed it?

[PRR]

> NE5532 and OPA2134 respectively) but then the volume boost (VR3 with a 100K poti) wouldn't work anymore.

'5532 has a lot of input bias current. A 1Meg bias resistor is sure to give tenths-Volt DC offset. Your boost network has DC gain. I can't say why it works at one spot. I can say I am not shocked.

'2134 is FET input, zero input current, so should not be a problem.

I suggest it is not build as diagrammed.

DC voltages can be your clues.

[lars-musik]

I suggest it is not build as diagrammed.

I used Diptrace to do a layout from that schematic, so at least the layout is identical to the schematic. I have to double check the components.


The circuits works now as advertised with the TL072. However, there is a new problem now.

The circuits needs about 3-4 seconds playing to fire up. When I plug everthing in it sound very quiet and distorted (even if I leave it a minute under power). After I hit the strings three times it is like an engine starting up: getting louder over the course of a few seconds and then it is completey there. There are no super-huge capacitors that need charging. What might cause such behaviour? Defect cap?

[PRR]

DC voltages can be your clues.

[lars-musik]

Allright. Now I got the message.

Here are the voltages from the blend circuit. I wonder if there is anything amiss with the charge pump. The output voltage seems a bit on the low side. I checked all traces under a stereo microscope with my DMM and all looks very well.

Do you spot something odd?

Thanks for your help!

[antonis]

Your DMM impedance dominates Input bias resistors..
(or else, you have 1μA input bias current..) 

What about U2.2 non-inverting Input..

[lars-musik]

Your DMM impedance dominates Input bias resistors..
(or else, you have 1μA input bias current..) 

Thanks antonis,

unfortunately I am more a hobbyist than I should be. Is there anything I could/should do about that? Just pretend I don't know anything about electronics  ......

[antonis]

Plain & simple reading for more hobbyists than they should be..

https://e2e.ti.com/blogs_/b/analogwire/archive/2016/12/08/what-you-need-to-know-about-input-bias-current-and-why
(you can skip complicated formulas..)

https://e2e.ti.com/blogs_/archives/b/thesignal/archive/2012/04/11/input-bias-current-cancelation-resistors-do-you-really-need-them

https://e2e.ti.com/blogs_/archives/b/thesignal/archive/2012/04/17/internal-input-bias-current-cancellation