Geofex Hum free splitter?

[simon111]

Just finished building the hum free splitter (3 way). Not sure if it's working correctly though. When two or three amps are connected then it's hum free (as long as one is connected to the direct output), but when using only one output (yes i know it's a splitter!) it hums like mad. Why's this? It does it only when I am running my little vox pathfinder on the output  and not my AC30. Also it hums worse when I am plugging the amplifier into a different mains socket than what I am running my splitter on (PSU being used).
Any help?

[R.G.]

Why's this?

There are some clues:

** When two or three amps are connected then it's hum free (as long as one is connected to the direct output)
** It does it only when I am running my little vox pathfinder on the output
** and not my AC30
** it hums worse when I am plugging the amplifier into a different mains socket than what I am running my splitter on (PSU being used).

When you don't use the direct output at all, then the whole circuit side is left completely without a ground reference, and you get hum like you would touching the end of an open cord. I think this could be cured by using a safety grounded power supply for the splitter, and then using NO direct outputs - but I haven't gone and done the tests to confirm that, and it removes some of the flexibility of using batteries. You have to have at least one ground reference on each side, I think.

Have you tried this with just batteries? It could also be that there is some small AC leakage from your PSU. That's pointed to by the notes that there are differences in amplifiers and where you plug in the PSU and amplifier.

I'd be the first to admit that the design is not a cure-all for all possible hum conditions. It was designed to fix hum when running one guitar into two or three amps - as you note, yes, it's a splitter. I'm reminded of the joke about a guy who walks into a doctor's office and says "Doc, it really hurts when I do ... this... " and the doctor's reply was "Well, don't do that." The simple answer is to plug something into the direct output if youu're only going to be using that one output. Plug secondary devices into the secondary outputs.

I think that your unit is working correctly - it just can't solve all possible problems.

[simon111]

RG.Cheers for that, I'll give the batteries a try, on a side note but still relevant, could you explain the need for the 11K and 10K + 2.2uF components at the pin2 connection? As my very poor understanding of opamps is that the 11K would add a slight gain to the circuit of 1.1 is this needed? why not connect pin2 to 6 for unity gain? and whats the 2.2uF whats that for? I am thinking about getting a no polarised 2.2uF ( I am using two 1uF electrolytics in parallel), would changing this help (i.e 2.2 over 2uF), or is it actually really needed at all? (must be a reason as to why it's there!)

Cheers.

[R.G.]

could you explain the need for the 11K and 10K + 2.2uF components at the pin2 connection? As my very poor understanding of opamps is that the 11K would add a slight gain to the circuit of 1.1 is this needed? why not connect pin2 to 6 for unity gain? and whats the 2.2uF whats that for?

You are correct, the components give a slight gain. This is there to compensate for the slight losses incurred in going through the transfomer. Real transformers always incur some loss in signal level because the windings have resistance. The signal is pre-boosted a bit to allow for that so the isolated signal comes out at about the same as the input signal. If I were doing this properly, I should have cut the buffered direct signal a bit to make it match. But the signal levels are very close, and most people prefer a slightly hotter signal if they can get it, even if they don't notice it as hotter.

whats the 2.2uF whats that for?

It blocks the unit from having a slight DC gain, and keeps the output centered on ground. This would not be a problem if the opamp was perfect, but there is an offset voltage in the opamp that would cause a slight DC offset. It's trivially small, but I put it in as good practice.

I am thinking about getting a no polarised 2.2uF ( I am using two 1uF electrolytics in parallel), would changing this help (i.e 2.2 over 2uF), or is it actually really needed at all? (must be a reason as to why it's there!)

There are strong reasons and weak reasons. It's there for a weak reason.

You would probably be OK to just leave it out and connect the 10K to ground. Likewise, you would probably be OK to leave the 10K *and* the 2.2uF NP out, just leaving the inverting input connected to the output through the 11K.

In fact, if you did that, you could probably just make the 11K be a dead short. This would give you a signal follower that would work pretty much the same but would have a small offset voltage of whatever the opamp does. It would work just fine like this for almost all opamps.

Intro to failures and warranty costs

In any design, there is always a set of choices to make on how much you put in to make the circuit behave politely, for want of a better term, and to be tough to possible abusive situations. In many cases, there are parts that are there just for the odd circumstances that might happen, but are not strictly necessary for sunny-day, good conditions operations. I wrote an article on that at http://geofex.com/circuits/what_are_all_those_parts_for.htm to explain some of these things. Designers usually toss in whatever protective and politeness parts they think are most likely to be needed, but rarely all of them unless it's a commercial design.

For instance, on this one, if I were making a commercial pedal, I would put a capacitor in series with the 10K at the input, in case someone put a 9V battery on the input. I would also put back-to-back diodes across the inputs, and reverse biased diodes from the input signal to the + and - power supplies in case someone hooked up a 110V line or a speaker output to the input. I would put a capacitor across the 11K resistor in the feedback loop to guard against oscillation if a capcitive cable was connected, a job that is shared by that series 10K resistor in the input. Strictly speaking, the 10K at the input and one of the 2.2M's are not needed either. That's for series protection of the input. I would probably also diode clamp or zener clamp - or both! - the output of the opamp to the power supplies to guard against someone putting 120Vac or a speaker output on the output jack. I'd likely put current limit clamps on the power supplies and reverse polarity protection on the power supplies as well.

The obvious question is - why??? If it works with fewer parts, why put more parts in? Doesn't that just make it a stodgy, stupid, expensive, military cost-overrun kind of design?

And the answer is the cost of fixing failures. If you have a commercial product, you can calculate pretty closely how often one of your boxes will fail just based on the parts list and the history of the failure rate of those parts, which is available information. You'll know, for example, that the failure rate of a 1N4148 signal diode is 0.00000037% per 1000 power on hours. What that tells you is that if you make 1000 pedals with a 1N4148 diode in each one, and these pedals each run for 1000 hours, then there is a 0.37% chance that ONE of them will have a failing diode. You can do that for all the parts in the pedal, adding up the probability of failure, and come up with a faiure rate for the total pedal.

Even though the individual failure rates are quite small, by the time you add up the whole parts list and multiply by the number of pedals you sell, you will come up with a very likely estimate of the number of failing pedals with time. And if you know how long your warranty is, you have a good estimate of how many pedals you are going to have to pay to replace or repair. That's the Intrinsic Failure Rate, or how often a pedal fails on its own.

But there are non-intrinsic failures too. If 10% of your buyers are bass players and regularly connect up their pedals to light sockets, speaker outputs, motor-generator sets, microwave ovens, and diesel tractors    , then you have a warranty repair problem. Your repair costs just went through the roof.

In this case, putting in cheap, low failure rate parts like resistors, caps and diodes to make the unit immune to many of the common misuses saves you far more than you would pay for the parts - and the increased failures the added parts put in. So in the commercial world, where repairs often cost a big fraction of the cost of the whole unit, avoiding even one repair pays for a lot of protective parts.

[simon111]

Great Read R.G. 

Thankyou so much for that. (already printed it out for safe keeping) - I am now going to try and modify mine using your 'commercial' ideas. That way I will have a circuit that I built myself that I KNOW is as good as can be.

Cheers
Simon.