Skip to content | Skip to navigation
Powered by RSPowered by RSPowered by RS

RF circuits-reg

9 May 2011

sreeyuktha

India

Hi, I am new to designing of circuits for RF. I intend to prepare a PCB for a circuit that is to amplify a signal of 1MHz frequency; the signal amplitude after amplification is about +/-450Volts.

I would like to know on following:

1. What are the specific points I need to look at from the High frequency and high voltage points of view in my circuit design, and PCB design?

2. Based on your experience, are there any standard Semiconductor chips for amplifiers that I can use for the above purpose? The input signal is a rectangular wave 30% duty-ratio, which is bipolar. I mean, it is a waveform with part of pulse (half cycle) on positive polarity and part of it in negtive polarity. Amplitude is about +/-2V.

Can someone help me out?

Replies

vincentcheung

New Zealand
1 year ago

Although the frequency 1MHz is not very high but with 450V amplitude, radiation can't be ignored.  A 30% duty cycle retangular wave is rich in both odd and even harmonics which means your signal contains significant amplitudes of 2MHz, 3MHz, 4MHz, 5MHz, 6MHz, ... as well.  If the rise and fall time of your signal are fast, the harmonics can easily go up up to VHF band.  So, the PCB must be designed to avoid interefences to AM, Short Wave, and FM broadcastings.

To design a low radiation PCB, the designer must know how the signal radiates.  For a frequency of 1MHz, the wavelength is 300 metre in space or about 200 metre on circuit boards which is much much longer than the dimension of a PCB.  So the radiation will mainly caused by current loops which act like loop antennae.  Since high voltage prevents you from routing high frequency signal traces close to their return paths for current loops minimization, the layout will be quite charlenging.  I recomend to use a perfect ground plane (no holes, no slots) for the high frequency areas.  Because of the high voltage, any traces next to the signal traces can cause radiating displacement currents flowing between the traces.  So, keep large distances between signal traces to all other traces or alternatively, run two thick ground traces with plenty vias to ground plane along both sides of every signal trace for shielding.  Of course there is an always correct rule - keep all signal traces as short as possible.  However, doing all about is not sufficient to make a PCB free from radiation because there may be some high frequency return currents not flowing through the return traces or groung plane adjacent to the signal traces which will cause net common mode rediating currents.  Techniques of avoiding common mode currents on PCBs are not simple enough to be discussed here.

In view of circuit design, one important point must be noted- always use toroide or ferrite beads to carry high RF currents, never use open end core coils because they are very good antennae.

Base on my experiences, I don't know any standard chips which can handle 1MHz square wave amplification up to +/-450V.  So, a design with discrete components may by unavoidable.  Depends on the requirements that the amplification needs to be linear low distortion or just keep signal timings with boosted voltage amplitudes, the design and components selection will be totally different.  For linear amplification of a DC-free signal (a +/-2V 30% duty cycle retangular wave input signal is not DC-free), a proper step up transformer may be used to avoid using high volt transistors.  For switching voltage boost, beware that switching 900V (from +450V to -450V and -450V to +450V) at 1MHz frequency can easily cause high radiating switching currents.  In this case, make use of circuit resonances to achieve zero voltage / zero current switching can greatly reduce radiation and increase power efficency.  Again, how to achieve ZVS and ZCS are not that simple too.

I hope this can help you a little bit!

 

 

jgmartinez

Spain
1 year ago

Well there are several questions to take in account.
1 Mhz is high frequency but not so high to take in account RF design technique. In my opinion not too many care has to be taken in account from the point of view of the frequency. Nevertheless in analog world is a good practice isolates as much as possible oscillator stage from output, input amplifier side from output, shielding if possible, avoid grounding circuits by a good ground plane, etc.
The second is the output voltage, 450 V. Well here are two actions, one is distance the other is PCB material. Separate as much as possible the high voltage points, perhaps some kind of potting will help, cover contacts or lugs with shrink material. PCB fiberglass probably is OK but it is better PTFE based PCB. It is very expensive but RF leakage is extremely low so with this material you’ll never have problems. If price OR PTFE availability is a problem, try before with the very common FR4 material and then check results to take a decision.