Power diode DSB modulator

[neazoi]

Hello, a very simple DSB (dual sideband) modulator can be made using a ring of diodes like this for example https://4.bp.blogspot.com/-feVxGR31GBU/UKeKlfEE4OI/AAAAAAAAS3c/t8E_JydAeg4/s1600/VK3YE+DSB.png or even better using a small transformer in the modulator like this **broken link removed**

I have made a simple but powerful CW keyed oscillator here **broken link removed**
I know that this is not the way things are done today, but I was wondering if I can drive a POWER diode DSB modulator with this, in combination with a POWER audio amplifier to produce a POWER DSB signal without the need for an RF amplifier.

 

[chuckey]

The problem is the frequency response of the cheapo 50 HZ rectifier diodes, they will have too high a stored charge to work efficiently at RF. So it depends on your working frequency and diode chosen.
Frank

 

[neazoi]

The problem is the frequency response of the cheapo 50 HZ rectifier diodes, they will have too high a stored charge to work efficiently at RF. So it depends on your working frequency and diode chosen.
Frank

The application is for HF 1-30MHz.
Are there any power schottky diodes for this frequency?
If not, I could possibly connect diodes in parallel to increase their current rating?

 

[betwixt]

The problem is stored charge as Frank points out. At low frequency the recovery time (how long it takes for the junction to change from being conductive to blocking the current) is insignificant but as the frequency increases it takes a larger proportion of each cycle time. Also bear in mind that the capacitance across the junction of a 1A power diode (1N400x for example) is typically around 30pF at 1V and for 3A diodes it is typically around 50pF. You would also need to find matched ones to make a good modulator.
Although it possibly isn't fair to compare their performance in low signal applications with high voltage/current applications, the highest frequency you could use a normal power diode is probably < 5KHz.

Power Schottky diodes would fare better because of their lower stored charge but I would still imagine their performance to be poor in signal mixing applications.

Brian.

 

[neazoi]

The problem is stored charge as Frank points out. At low frequency the recovery time (how long it takes for the junction to change from being conductive to blocking the current) is insignificant but as the frequency increases it takes a larger proportion of each cycle time. Also bear in mind that the capacitance across the junction of a 1A power diode (1N400x for example) is typically around 30pF at 1V and for 3A diodes it is typically around 50pF. You would also need to find matched ones to make a good modulator.
Although it possibly isn't fair to compare their performance in low signal applications with high voltage/current applications, the highest frequency you could use a normal power diode is probably < 5KHz.

Power Schottky diodes would fare better because of their lower stored charge but I would still imagine their performance to be poor in signal mixing applications.

Brian.

It would be interesting to try this out though, I guess...

Is there really a power Schottky diode that can is intended for RF use up to 30MHz?
Maybe I could try paralleling smaller diodes but if they are not matched, some of them may switch sooner than others, with unknown output effect.

I guess that any fast diode can be used, it only needs to be a power one, but how much power, is a question.
This transmitter outputs a max of 8W at 50R so this is about 57Vpp.

What sort of parameters should I look for in the datasheets?

Please see the attached datasheet, could this be an option?

 

[Dan Mills]

Remember that for a classical diode mixer, the LO power is usually very much greater then the power at either of the other ports. This is to ensure that the diodes are fully switched (For example a level 17 mixer (+17dBm LO drive, might work at rf/if levels below 0dBm).

What might be interesting is to try using a couple of power diodes (that often have effectively a PIN structure) as variable attenuators, with the high voltage side being the modulating envelope, would probably need considerable bias but might be OK as you are now looking for some part with a carrier lifetime greater then an RF cycle but smaller then the period of the modulating signal.

Mixers, fun for all the family....

Regards, Dan.

 

[neazoi]

Remember that for a classical diode mixer, the LO power is usually very much greater then the power at either of the other ports. This is to ensure that the diodes are fully switched (For example a level 17 mixer (+17dBm LO drive, might work at rf/if levels below 0dBm).

What might be interesting is to try using a couple of power diodes (that often have effectively a PIN structure) as variable attenuators, with the high voltage side being the modulating envelope, would probably need considerable bias but might be OK as you are now looking for some part with a carrier lifetime greater then an RF cycle but smaller then the period of the modulating signal.

Mixers, fun for all the family....

Regards, Dan.

Indeed this is the case. What I am trying to do is a kind of special, in the sense that I am trying to mix (modulate) two high power signals (RF LO and Audio) using power diodes, or else a power diode ring mixer. This is not the way things are done but I am doing it as an experiment, hoping to produce a high power DSB signal directly.

I saw the PIN diodes you refer to and especially the power PIN diodes, but then I found the BYV26xx diodes attached in previous post. With 8-35pF capacitance they do not seem that bad for say 7MHz, but I may be reading something wrong, that is why I am asking.

 

[neazoi]

The problem is stored charge as Frank points out. At low frequency the recovery time (how long it takes for the junction to change from being conductive to blocking the current) is insignificant but as the frequency increases it takes a larger proportion of each cycle time. Also bear in mind that the capacitance across the junction of a 1A power diode (1N400x for example) is typically around 30pF at 1V and for 3A diodes it is typically around 50pF. You would also need to find matched ones to make a good modulator.
Although it possibly isn't fair to compare their performance in low signal applications with high voltage/current applications, the highest frequency you could use a normal power diode is probably < 5KHz.

Power Schottky diodes would fare better because of their lower stored charge but I would still imagine their performance to be poor in signal mixing applications.

Brian.

Hi, I am sorry to open this relatively old topic again,
I was wondering if the BYV26 can do the job as a power dsb modulator for HF?
If yes, then which is more important for the choice, the lowest reverse recovery time or the diode capacitance?

 

[betwixt]

I think the only way to know for sure is to try it. I suspect you will be disappointed by the results though.
You would be using them for a purpose far away from what they were designed for so it's anyones guess as to how well they may work. The capacitance is important, lower is better but I'm not sure how the soft recovery characteristics will influence performance at very high frequencies.

Brian.

 

[neazoi]

I think the only way to know for sure is to try it. I suspect you will be disappointed by the results though.
You would be using them for a purpose far away from what they were designed for so it's anyones guess as to how well they may work. The capacitance is important, lower is better but I'm not sure how the soft recovery characteristics will influence performance at very high frequencies.

Brian.

These diodes are about 35pF but this is the maximum if you see the graph, and the reverse recovery time is about 30ns.

I wonder if I can put two diodes in series to reduce the overal capacitance.

I also wonder if I can put smaller (but faster) diodes in parallel to create a composite power diode. For example the 1n5712 has only 1.2pf capacirance, so I can put several of them in parallel without increasing the total capacitance in unacceptable levels for HF?

- - - Updated - - -

Hey, I just got another idea, instead of using diodes, why not using the B-E junction of an RF power transistor?
2sc2166 are really cheap!

 

[betwixt]

I've seen bicycle wheels with window blinds clipped to the spokes to make wind turbines - but if you asked the bicycle manufacturer how much power it would produce they wouldn't be able to tell you.

Same with power diodes used for purposes they were not designed for, they would tell you to use an appropriate device or test it for yourself. The recovery time isn't a fixed amount, it depends upon the step in voltage it is recovering from and to some extent how much of the stored charge is leaked away from outside the device. If you take the worst case 150nS recovery figure for the BYV26 for example, it would suggest it would totally be unable to conduct at frequencies as low as 6.6MHz and performance would be degraded as that frequency was approached. Typical use for that device would be in SMPS at ~150KHz or less.

Connecting two diodes in series would reduce the capacitance but also double the power loss and voltage necessary to make them forward conduct. Connecting them in parallel, unless you were very careful to match them, would result in unequal currents flowing through them.

Consider also that if you are trying to make DSB at high power levels, you are likely to produce significant harmonic output which would reduce efficiency. You can't just connect an antenna after the diodes, apart from the interference risks, you would probably have to adjust the balance at each frequency to account for changes to the antenna resonance. It is generally better and easier to generate the signal at low level, filter it and then amplify it. That also gives you the opportunity to remove one sideband if you want SSB.

Brian.

 

[neazoi]

I've seen bicycle wheels with window blinds clipped to the spokes to make wind turbines - but if you asked the bicycle manufacturer how much power it would produce they wouldn't be able to tell you.

Same with power diodes used for purposes they were not designed for, they would tell you to use an appropriate device or test it for yourself. The recovery time isn't a fixed amount, it depends upon the step in voltage it is recovering from and to some extent how much of the stored charge is leaked away from outside the device. If you take the worst case 150nS recovery figure for the BYV26 for example, it would suggest it would totally be unable to conduct at frequencies as low as 6.6MHz and performance would be degraded as that frequency was approached. Typical use for that device would be in SMPS at ~150KHz or less.

Connecting two diodes in series would reduce the capacitance but also double the power loss and voltage necessary to make them forward conduct. Connecting them in parallel, unless you were very careful to match them, would result in unequal currents flowing through them.

Consider also that if you are trying to make DSB at high power levels, you are likely to produce significant harmonic output which would reduce efficiency. You can't just connect an antenna after the diodes, apart from the interference risks, you would probably have to adjust the balance at each frequency to account for changes to the antenna resonance. It is generally better and easier to generate the signal at low level, filter it and then amplify it. That also gives you the opportunity to remove one sideband if you want SSB.

Brian.

Thaks Brian,
I really understand how bad this mixer would be for the purpose, no doubt.

However why not experimenting to see the results? How about a single balanced mixer but replacing the diodes with something like this https://i.stack.imgur.com/D8oHk.png to create a "power" fast diode.
Or maybe using a power transistor like 2sc2166 and connect the collector with the base, to create a "power" fast diode?

Look at what I have found on the net:

"It is a common practice in precision analog electronics to use Bipolar transistors as diodes. The purpose is to obtain a very low leakage diode. For example, a 3904 type transistor will have <1pA of reverse leakage using the Base Emitter junction. However, it turns into a zener diode at around 6.8V. Works great for 5V and lower voltage logic circuits. Higher current and reverse voltage is achieved by using the Base as anode and the collector as the cathode. Still an excellent low leakage diode at around 10pA and now you get the voltage rating of the transistor and improved current. This will not be a high speed diode. Higher speed is achieved by shorting Collector to Base (Anode) and using the emitter as Cathode. However, reverse voltage must be limited to <5V."

 

[betwixt]

The image doesn't make sense, MOSFETs like the IRF530 have built in avalanche diodes between the drain and source pins but connecting the gate to the source just stops the transistor part of it ever working. The result is something like normal power diode (1N540x for example) but with extra capacitance across it.

The bipolar transistor with base and collector linked will work but the reverse breakdown voltage of the B-E junction will seriously limit the voltage you can use it at. The 2SC2166 is a good HF amplifier, I've used lots of them in the past but the rated breakdown voltage is only 5V. When you factor in the voltage from the power oscillator and voltages from the output circuit, which may be out of phase and therefore additive, you will likely overload it at quite low RF power.

Brian.