Why LEDs create higher sound volume compared with other diodes when used in Ring modulator

Fluffyboii said:

I failed to find anyone selling 1:1 transformers locally.

Click to expand...

Audio transformers for telephone modems are usually 1:1 and pretty good at audio as the modems needed quite wide bandwidth with good frequency response and linearity.

See if you can find some old 56K (V92) modems for people who have not thrown them out?

A typical example on ebay, the transformer is obvious:

You would need two identical transformers at either end with the windings in series to simulate a centre tap, as they are usually single winding each side.

You can also get cheap generic 1:1 audio transformers on ebay, eg:

Or with centre taps, so only one needed each side, but a bit more expensive:


I'm not sure about the ring mod diode, I'd have to see the full circuit you are using for the modulation drive and transformer & diode configuration.

I'd expect the carrier input to be current controlled, eg. with a series resistor to regulate the diode current in proportion to voltage. The transformer characteristics will also have a massive effect.

If the modulation drive is too low in proportion to the signal level, the signal current could "switch off" diodes that are supposed to be conducting due to carrier current, or even switch on ones that are supposed to be off.

Using LEDs would mean a far higher signal level was needed to cause false conduction, but if everything is properly set up such things should not really be a problem?

rjenkinsgb said:

I'd have to see the full circuit you are using for the modulation drive and transformer & diode configuration.

Click to expand...

I am currently only using the common 6 parts passive circuit. I use an dual op amp with around 10 times gain for inreasing and buffering my mic preamp gain and decrease the 12V square wave with a 100K pot at the input.

The transformers as I said are custom made with 3400 turns of 0.07mm wire with middle tap on other side at 1700th turn. It measures 750 ohms at middle tapless side and measures 300-400 ohms on other side. It is a double section type of core plastic as you can see in image. These cores are standart I think, I could not find bigger ones that had seperated sections for primary and secondary. I could just wind them over each other like commercial ones but decided to use the cores custom transformer shop recommended me.

I had to go 3400 turns because there was no way 800-1000 turns of thicker wire would give me 600ohm resistance like those ebay ones. In fact using 0.18 wire at 1000 turns only had arouns 70 ohms so it had to be rewound. Idk if those Ebay transformers use even thinner wire or the base is so thick that perimeter is too large and the wire length is too high. I was lucky to find a shop owner that was willing to spend time on such a small offer, normally they don't bother with one or two transformers of this size.

Maybe I should add two input op amps in the given circuit for adjusting input volumes as well. Actually I think it makes more sense to amplify input instead of output to have the circuit operating at LEDs sweet spot.

LEDs work much better because they have exact same voltage drop and I think it is because they have hifher voltage drop that small differences between them doesn't matter. Germanium ones have 0.633Vd or 0.730Vd changing between same samples so I can hear carrier when I use them since it is not balanced. I need ones with exact Vd, I will take my test device at electronics shop to do that later. I am glad that I put female pin holders on my circuit so I can test all kinds of diodes without desoldering.

Fluffyboii said:

The transformers as I said are custom made with 3400 turns of 0.07mm wire with middle tap on other side at 1700th turn. It measures 750 ohms at middle tapless side and measures 300-400 ohms on other side. It is a double section type of core plastic as you can see in image. These cores are standart I think, I could not find bigger ones that had seperated sections for primary and secondary. I could just wind them over each other like commercial ones but decided to use the cores custom transformer shop recommended me.

I had to go 3400 turns because there was no way 800-1000 turns of thicker wire would give me 600ohm resistance like those ebay ones. In fact using 0.18 wire at 1000 turns only had arouns 70 ohms so it had to be rewound. Idk if those Ebay transformers use even thinner wire or the base is so thick that perimeter is too large and the wire length is too high. I was lucky to find a shop owner that was willing to spend time on such a small offer, normally they don't bother with one or two transformers of this size.

Click to expand...

I suspect you're a little confused about resistance and impedance?, I presume the Ebay ones will be 600 ohm IMPEDANCE, and a far lower resistance than that.

This is a brief spec. from a VERY (you don't want to look!) expensive 600 to 2.4K transformer from RS Components:

Primary Impedance	150 Ω, 600 Ω
Secondary Impedance	2.4 kΩ, 600 Ω
Primary DC Resistance	15 (Primary 1) Ω, 15 (Primary 2) Ω
Secondary DC Resistance	74.2 (Secondary 1 and 2)Ω

Click to expand...

Notice how low the DC resistances are. The actual turns ratios are:

Turns Ratio

1+1:2+2

Click to expand...

Essentially you don't wind a transformer to get the impedance you want, it's basically just a function of what impedances you connect to it - so if you have a 1:1 transformer with a 600 ohm load on it, then the primary impedance will be 600 ohms. In you then make the load 6000 ohms, then the primary impedance will be 6000 ohms.

Obviously there are restrictions in how far this works, and you need a certain number of turns in order to achieve your desired result, but that should then work over a pretty wide range. Obviously if you only had one turn on the primary, and one turn on the secondary, then it wouldn't work terribly well for an audio transformer - but would be fine for a high frequency RF transformer. Likewise, while your 3400 turn transformer works OK at audio, it would be worse than useless at high frequency RF.

Nigel Goodwin said:

I suspect you're a little confused about resistance and impedance?, I presume the Ebay ones will be 600 ohm IMPEDANCE, and a far lower resistance than that.

This is a brief spec. from a VERY (you don't want to look!) expensive 600 to 2.4K transformer from RS Components:



Notice how low the DC resistances are. The actual turns ratios are:



Essentially you don't wind a transformer to get the impedance you want, it's basically just a function of what impedances you connect to it - so if you have a 1:1 transformer with a 600 ohm load on it, then the primary impedance will be 600 ohms. In you then make the load 6000 ohms, then the primary impedance will be 6000 ohms.

Obviously there are restrictions in how far this works, and you need a certain number of turns in order to achieve your desired result, but that should then work over a pretty wide range. Obviously if you only had one turn on the primary, and one turn on the secondary, then it wouldn't work terribly well for an audio transformer - but would be fine for a high frequency RF transformer. Likewise, while your 3400 turn transformer works OK at audio, it would be worse than useless at high frequency RF.

Click to expand...

I assumed 600ohm would be DC resistance and 10K ohm would be impedance for transformers people use for ring modulators. People use like "20K 1:1 transformers" for having full human hearing range. I thought by having a high turn number and high coil resistance I would have a transformer that does not load down the input source and have better performance when used in a ring modulator.
My component tester tells me that my transformers have 750 ohm DC resistance as expected and 26 Henry inductance with metal core in it. I am very confused about transformers indeed.

Added a typical transformer people use for ring modulators.

Fluffyboii said:

I assumed 600ohm would be DC resistance and 10K ohm would be impedance for transformers people use for ring modulators. People use like "20K 1:1 transformers" for having full human hearing range.

Click to expand...

No, it's certainly NOT DC resistance, and I would suggest the 20K you mention is possibly the suggested frequency range? - up to 20KHz.

Fluffyboii said:

I thought by having a high turn number and high coil resistance I would have a transformer that does not load down the input source and have better performance when used in a ring modulator.

Click to expand...

Sorry, but you thought wrongly - your initial assumption of 600 ohms being resistance, and not impedance, has led you astray.

You mentioned "cores custom transformer shop", didn't they advise you? - or didn't you actually tell them what you were trying to do?.

Fluffyboii said:

My component tester tells me that my transformers have 750 ohm DC resistance as expected and 26 Henry inductance with metal core in it. I am very confused about transformers indeed.

Click to expand...

Notice your 750 ohms DC resistance is wildly different to the 15+15 ohm DC resistance of the professional transformer.

Do you have a scope and an audio frequency generator?, it might be interesting to test your transformer over it's full frequency range, with different loads.

Fluffyboii said:

Added a typical transformer people use for ring modulators.

Click to expand...

The LT44 was a VERY old driver transformer, basically used back before they discovered how to make transistor amplifiers There was a second used as an output transformer (I can't remember it's Eagle number?), which fed from the output transistors to the speaker - it was rather like a VERY old valve design. Here's a classic old circuit - the 'D' in OC81D stands for 'driver'.

Nigel Goodwin said:

No, it's certainly NOT DC resistance, and I would suggest the 20K you mention is possibly the suggested frequency range? - up to 20KHz.



Sorry, but you thought wrongly - your initial assumption of 600 ohms being resistance, and not impedance, has led you astray.

You mentioned "cores custom transformer shop", didn't they advise you? - or didn't you actually tell them what you were trying to do?.



Notice your 750 ohms DC resistance is wildly different to the 15+15 ohm DC resistance of the professional transformer.

Do you have a scope and an audio frequency generator?, it might be interesting to test your transformer over it's full frequency range, with different loads.


The LT44 was a VERY old driver transformer, basically used back before they discovered how to make transistor amplifiers There was a second used as an output transformer (I can't remember it's Eagle number?), which fed from the output transistors to the speaker - it was rather like a VERY old valve design. Here's a classic old circuit - the 'D' in OC81D stands for 'driver'.

View attachment 138344

Click to expand...

I do have a scope and I can feed a saw or square wave into the transformer with it and see how it responds. Average transformer winding shop doesn't really know about electronics so I said I would need around 600 turns and they gave me the cores and 0.45mm wire for it. Obviously 0.45 was to thick so I found another shop and get them done with 0.07mm wire. Since I will be using a buffer at this things output I assume it would be a very high resistance to simulate that. Lemme try it out and see what is the bandwidth. Honestly this sounds fine so I don't think I will go and got more transformers done.

Nigel Goodwin said:

Do you have a scope and an audio frequency generator?, it might be interesting to test your transformer over it's full frequency range, with different loads.

Click to expand...

It looks like after 10Khz the waves start getting all wiggly from harmonics or resonance. It actually looks like what a resonance circuit does to a waveform. The square wave starts turning into a sine wave and I get a sine wave after 12Khz. The volume starts decreasing after not much higher than that and almost fully silences when I reach 20Khz.

Edit: It looks like my square wave generator gives up after 15Khz and square wave stops being square wave without the transformer. So that data is useless. But it really adds some resonance to sound, that part is true. I will build a 555 timer circuit and use that.

Fluffyboii said:

Average transformer winding shop doesn't really know about electronics so I said I would need around 600 turns and they gave me the cores and 0.45mm wire for it.

Click to expand...

They don't need to know about electronics, but presumably they do know about transformers?.

If you told them you were wanting to wind an audio isolation transformer, for 600 ohms, they would presumably (and hopefully) made suitable suggestions as to the core, wire size, and number of turns.

Nigel Goodwin said:

They don't need to know about electronics, but presumably they do know about transformers?.

If you told them you were wanting to wind an audio isolation transformer, for 600 ohms, they would presumably (and hopefully) made suitable suggestions as to the core, wire size, and number of turns.

Click to expand...

They don't really know how to calculate resistance. You tell them number of turns and wire thickness and they wind it for you. Thats why we tried 3 different wire thicknesses before using 0.07mm wire to find out what resistance would be. The first shop I went have me 0.45mm wire which only fit 120 turns on the core. He gave it to fit would fit 600. So yeah not the most educated people around the world.

Fluffyboii said:

It looks like after 10Khz the waves start getting all wiggly from harmonics or resonance. It actually looks like what a resonance circuit does to a waveform. The square wave starts turning into a sine wave and I get a sine wave after 12Khz. The volume starts decreasing after not much higher than that and almost fully silences when I reach 20Khz.

Edit: It looks like my square wave generator gives up after 15Khz and square wave stops being square wave without the transformer. So that data is useless. But it really adds some resonance to sound, that part is true. I will build a 555 timer circuit and use that.

Click to expand...

You really need a sine wave, and a 555 wouldn't make a very good audio generator.

You can simply use your PC's sound card to generate your sine wanes, here is alist of suitable software:

Fluffyboii said:

They don't really know how to calculate resistance. You tell them number of turns and wire thickness and they wind it for you. Thats why we tried 3 different wire thicknesses before using 0.07mm wire to find out what resistance would be. The first shop I went have me 0.45mm wire which only fit 120 turns on the core. He gave it to fit would fit 600. So yeah not the most educated people around the world.

Click to expand...

Why would they know how to calculate resistance?, it's not a transformer requirement - for audio transformers impedance is. However, if they only wind power transformers?, they may not know much about audio transformers.

Interestingly it converts square into a sine wave after 3khz or so. You can see the peak at 12Khz. It is very interesting. And I read a higher voltage at middle tap even though at there it should act like a step down transformer.

I am getting very inconsistent results. I removed the diodes from the circuit and tested one transformer that way but maybe I should fully remove it. I just don't understand anymore. Since it sounds fine I think I will keep it as it is. I don't want to go and make a 600 turn one. I wish I asked this before getting the transformers.

Fluffyboii said:

Interestingly it converts square into a sine wave after 3khz or so. You can see the peak at 12Khz. It is very interesting. And I read a higher voltage at middle tap even though at there it should act like a step down transformer.

Click to expand...

A square wave is effectively made of odd harmonics of sine waves, so as the higher frequencies roll off the output waveform becomes more sinusoidal - until eventually you're left with just the fundamental sine wave.

Presumably this shows that your transformer doesn't handle high frequencies very well? - I would expect this to perhaps be a sign of far too many windings used?.

Nigel Goodwin said:

A square wave is effectively made of odd harmonics of sine waves, so as the higher frequencies roll off the output waveform becomes more sinusoidal - until eventually you're left with just the fundamental sine wave.

Presumably this shows that your transformer doesn't handle high frequencies very well? - I would expect this to perhaps be a sign of far too many windings used?.

Click to expand...

Thats probably correct. I will test the sine wave now.

Fluffyboii said:

I don't really understand why LED's produce high volume output. They should decrease volume more because they have 1.84V voltage drop in this case while Germanium diodes have around 0.7V voltage drop. Yet Germanium diodes are quieter.

Click to expand...

The diodes are only supposed to be acting as switches, with one pair or the other linking the signal between the two transformers, depending on the carrier polarity.

If the carrier is under-driver or the signal over-driven, the "on" diodes will start to be turned off by the audio signal.

Small signal or schottky diodes turn off very fast, while normal LEDs can be relatively slow - that may be the reason for the effect you are seeing, the LED switching speed is compensating for deficiencies elsewhere in the circuit.

It could be interesting to try some 1N4007 rectifier diodes, as they also have very slow reverse recovery (turn-off) in comparison to signal diodes?

rjenkinsgb said:

The diodes are only supposed to be acting as switches, with one pair or the other linking the signal between the two transformers, depending on the carrier polarity.

If the carrier is under-driver or the signal over-driven, the "on" diodes will start to be turned off by the audio signal.

Small signal or schottky diodes turn off very fast, while normal LEDs can be relatively slow - that may be the reason for the effect you are seeing, the LED switching speed is compensating for deficiencies elsewhere in the circuit.

It could be interesting to try some 1N4007 rectifier diodes, as they also have very slow reverse recovery (turn-off) in comparison to signal diodes?

Click to expand...

Ok with the sine wave things are much clearer. The transformers shows 1:1 characteristics from 100Hz to 10Khz.
The voltage starts increasing when I get closer to 15K. It increases from 1V input to 5V output at peak for reasons I can't understand. After 16Khz it starts dropping back to normal. Around 20Khz it drops the same voltage with the input again and starts decreasing rapidly from there. I think the bandwidth is fine but there is some interesting stuff going on with these transformers.

I should have 4007 diodes at somewhere. Should I make the same setup as I did at the start with mic preamplifier and square wave around 30Hz.

Fluffyboii said:

The voltage starts increasing when I get closer to 15K. It increases from 1V input to 5V output at peak for reasons I can't understand. After 16Khz it starts dropping back to normal. Around 20Khz it drops the same voltage with the input again and starts decreasing rapidly from there. I think the bandwidth is fine but there is some interesting stuff going on with these transformers.

Click to expand...

Perhaps you're hitting the resonant frequency of the transformer?. Adding a capacitor across the transformer should shift the peak lower in frequency. With no capacitor at all it's resonant with it's own 'self capacitance' and any 'stray capacitance' that's around. Try a 0.01uF across the transformer, and see how that effects the peak.

Nigel Goodwin said:

Perhaps you're hitting the resonant frequency of the transformer?. Adding a capacitor across the transformer should shift the peak lower in frequency. With no capacitor at all it's resonant with it's own 'self capacitance' and any 'stray capacitance' that's around. Try a 0.01uF across the transformer, and see how that effects the peak.

Click to expand...

Will try that and report back. But seeing the bandwidth very wide makes me think these are not unusable after all, what you think?

I realised that I forgot to add a load as well. Since I will buffer it with an op amp the resistor that is needed to emulate that would need to be very high right? Shall I try a megaohm or 100K or something between perhaps. Something like 1K load would drop the voltage a lot since the coil wire is very thin I assume.

Fluffyboii said:

Will try that and report back. But seeing the bandwidth very wide makes me think these are not unusable after all, what you think?

I realised that I forgot to add a load as well. Since I will buffer it with an op amp the resistor that is needed to emulate that would need to be very high right? Shall I try a megaohm or 100K or something between perhaps. Something like 1K load would drop the voltage a lot since the coil wire is very thin I assume.

Click to expand...

The load shown in your circuit of post #3 is 10K, the load should be whatever you wanted the output impedance of the transformer to be. Lowering the load resistance also lowers the input impedance of the transformer. A load resistor should also flatten your peak out somewhat, and potentially might improve your bandwidth as well?.