High speed sine wave to square wave converter

[thavamaran]

Hi guys, I need to convert a sine wave to square wave, sine wave is running at about 1.5 GHz, so I had an idea of using comparator, but I am afraid that most of the conventional op-amps can't support my speed requirement.

Any advice and reference circuits please.

 

[berkeman]

Hi guys, I need to convert a sine wave to square wave, sine wave is running at about 1.5 GHz, so I had an idea of using comparator, but I am afraid that most of the conventional op-amps can't support my speed requirement.

Any advice and reference circuits please.

That's definitely high speed. What is the source of the 1.5GHz "square wave"? What are the rise and fall times? Is the source impedance 50 Ohms?

What are you going to do with the "sine wave"? What THD can you tolerate? What is the load impedance? Can you just use a bandpass filter at 1.5GHz?

 

[thavamaran]

Hi berkeman, the description that you are giving sounds like square to sine. What I need is sine to square, I am using an Agilent vector signal generator to generate sine wave at 1.5 GHz, so I need to convert this to square 1.5 GHz.

 

[berkeman]

Hi berkeman, the description that you are giving sounds like square to sine. What I need is sine to square, I am using an Agilent vector signal generator to generate sine wave at 1.5 GHz, so I need to convert this to square 1.5 GHz.

Ah, apologies for my misread.

So, what are you going to do with the square-ish wave? What edge speeds do you need? What amplitude sine are you starting with, and what amplitude square wave do you want? If you can start with enough amplitude on your sine wave, you could use diode clamps to flatten off the tops...

 

[schip666!]

Is there any logic family that works at that frequency these days? There must be, but I haven't been following the industry. If so, and your sine wave voltage was appropriate, you could probably just use an inverter or buffer device. For that matter an RF transistor in a common-emitter config might just do the trick...

 

[thavamaran]

berkeman, no worries, thanks a lot for replying...

Im going to use this square-ish wave to function as battery for biasing device. I actually need a rectangular pulse with on-time 1ns and off time 1ns which the frequency is 500 MHz, so I just explain as 1.5 GHz cause in future I might increase the speed.

Sine wave amplitude is 1.8V and can i get the same for the rectangular? Diode clamps means diode clipping circuit is it? But will diode work this fast?

 

[thavamaran]

schip666!, there is this schmitt trigger which works above 1 GHz (SN74LVC1G17DB). Inverter or buffer?

 

[skeptic2]

Think about it in the frequency domain. In order to get anything close to a square wave you're going to have to create a significant number of odd harmonics. So don't think so much about 1.5 GHz but about maybe 9 GHz. I don't know of any logic or comparators that can do that (they may exist but they'll be expensive).

I'd get some microwave transistors and amplify the sinewave into clipping until I reach the maximum slew rate of the transistors. Using more stages instead of fewer will probably get you a better squarewave and try not to drive the stages too much into saturation or cutoff.

 

[sophiecentaur]

berkeman, no worries, thanks a lot for replying...

Im going to use this square-ish wave to function as battery for biasing device. I actually need a rectangular pulse with on-time 1ns and off time 1ns which the frequency is 500 MHz, so I just explain as 1.5 GHz cause in future I might increase the speed.

Sine wave amplitude is 1.8V and can i get the same for the rectangular? Diode clamps means diode clipping circuit is it? But will diode work this fast?

It still isn't clear it needs to be a square wave. If you are 'biasing' ("battery") that implies a constant value of voltage. If you are 'switching' / modulating / sampling then the harmonics associated with a waveform that is 'squarer' than your original 'sine wave' will be well out of band.

You may get a lot of irrelevant answers if you aren't a bit more detailed about your application. I have a feeling that the useful answer may be that you may not need to have a square wave.

 

[Studiot]

I think skeptic's overdriven amplifier/switch approach is the likely to be the most fruitful.

Try this device

http://uk.rs-online.com/web/search/searchBrowseAction.html?method=getProduct&R=4969212

As pointed out to get any sort of square wave you need a bandwidth of 3 - 5 times the fundamental. Otherwise you will just get a rounded triangular wave.

I suggest you try this at lower frequency to see what we mean.

 

[Mike_In_Plano]

Making a transistor amp that saturates and comes back out in a few 100 ps sounds pretty crazy. Diodes are far faster and can be saturated, amplified, saturated again, amplified again... There's no such thing as a "clean" square wave at 1.5GHz, but this would get you something without undue pain.

There are also GaAs houses producing flip-flop dividers over 10GHz. If you have source, those could be fairly clean.

Finally, you might look into onsemi's latest generation of ECL. Based on SiGe, these can give good logic edges.

- Mike

 

[sophiecentaur]

To make any really relevant comment about this is it really necessary to know a bit more about the actual application. I might ask whether this question relates to a practical situation or a simulated one. It's all very well finding a device that will produce this "square wave" when operating into a resistive load of the right value but it's an entirely different matter to say what it's going to look like when fed into a base. I have to ask again, what it is actually needed for and what is the specific need for a square wave.

 

[skeptic2]

Mike_In_Plano that's a pretty good idea. You'd still need several stages of amplification using a higher voltage supply before the diode clipper but the diode clipper would probably give a cleaner square wave.

 

[sophiecentaur]

Mike_In_Plano that's a pretty good idea. You'd still need several stages of amplification using a higher voltage supply before the diode clipper but the diode clipper would probably give a cleaner square wave.

... when applied to what load?

 

[Studiot]

Looked again at my reference, somehow got the wrong one. It should have been a 45Ghz MMIC switch.

http://eprints.ecs.soton.ac.uk/17613/1/switch_paper_090616.pdf

Sorry.

This is extreme, but there are plenty of MMIC devices on the market to satisfy your needs.