Increase Bandwidth of DIY Comb Generator

[darthhideous]

I need a comb generator that can operate between 1MHz -> 1GHz with 10MHz spacing, so I made a very basic one with an oscillator, cap and a diode. When I tested it, the spikes die off at around 300MHz. Does anyone have suggestions on increasing the bandwidth out to 1GHz? I've been looking into frequency mixing but not entirely sure if this will work as expected.

[awallin]

could you just take a 10MHz source of suitable stability, pass it through a high-speed logic-gate that produces square wave, and use that either directly or do some additional filtering to get pulses/spikes at each transition?

it should be fairly easy to find cheap logic-gates with 500 ps or so rise/fall times, which should give harmonics up to 1 GHz (?)

[mmagin]

I would expect your problems lie in the PC board layout or the choice of diode.  How about a photo of the board and what part number you used for the diode?

[David Hess]

Varactor diodes are constructed similarly to step recovery diodes so you might try some of them assuming you can find ones with a reverse recovery time shorter than the 50 nanoseconds from your 10MHz source.

[IconicPCB]

Try something like a high Ft transistor and overdrive it. Say a BFR92 or something similar.

[David Hess]

Try something like a high Ft transistor and overdrive it. Say a BFR92 or something similar.

Oh, that reminds me.  Some transistor junctions make excellent step recovery diodes.

[tggzzz]

I need a comb generator that can operate between 1MHz -> 1GHz with 10MHz spacing, so I made a very basic one with an oscillator, cap and a diode. When I tested it, the spikes die off at around 300MHz. Does anyone have suggestions on increasing the bandwidth out to 1GHz? I've been looking into frequency mixing but not entirely sure if this will work as expected.

At these frequencies the devil is in the details. 1mm of wire =1nH. Work out the ^di/_dt and then calculate the induced voltage.

To get helpful responses, you need to post a picture of the schematic and of the implementation of the schematic.

[jungle vegetable]

The easiest way is probably using a digital circuit with output rise time in the range you want (i.e. 200ps or less). The choice of digital circuits would be pretty wide, e.g. logic gates, comparators, clock buffers, limiting amplifiers and more. CML outputs would be preferable for easy 50 ohm output matching

I personally had great success with limiting amplifiers like ONET8501P - they have differential inputs with high sensitivity and 50 ohm CML outputs with fast rise times and little or no ringing

[Bart]

In this paper they use a PECL-Flipflop (MC100EP31), which is reset by the own output:
http://jpier.org/PIERL/pierl70/10.17070403.pdf

So you get a short pulse with the length of the propagation delay.

I've build a copy of this circut (with more SMD parts) and get similar results.

regards,
Bart

[tggzzz]

I need a comb generator that can operate between 1MHz -> 1GHz with 10MHz spacing, so I made a very basic one with an oscillator, cap and a diode. When I tested it, the spikes die off at around 300MHz. Does anyone have suggestions on increasing the bandwidth out to 1GHz? I've been looking into frequency mixing but not entirely sure if this will work as expected.

Just use a modern fast logic gate. The 74lvc family can have <300ps risetime, which translates to 1GHz.

https://www.eevblog.com/forum/testgear/show-us-your-square-wave/msg1902941/#msg1902941

If you don't like the LF content, insert a high pass filter.

[KE5FX]

Some notes from a few years back on the subject, with references for further reading. 

TL,DR: overdriven MMIC amps are good for making quick and (not so) dirty comb generators.  The best results I saw at the time came from cascading multiple stages with very low-value coupling capacitors.