Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

H-bridge / Breadboard noise

Status
Not open for further replies.

jrz126

Active Member
I'm currently building an inverted pendulum for a final project and I'm running into some problems.

There is aton of noise on all of the voltage supplies from the pwm switching and it is making it difficult to measure some rotary encoder signals.

I tried using optoisolators, but my encoder signals are too fast to use them.
(the pendulum encoder has a pulse width of 5us).

Any suggestions on how to remove some of this noise?

I'm using these chips https://www.electro-tech-online.com/custompdfs/2006/12/ir2184.pdf to drive the bridge and it says that they have a seperate power and logic ground, but I cant seem to get it to drive the highside fet with the grounds seperate. Anyone familar with using this IC?
 
Do you have enough bypass capacitors on the supply lines?

I'm not familiar with the IC you mentioned, but in the case of A/D converters, the manufacturer recommends that the connection between the analogue and digital ground planes be under the A/D package and be no more than 3/8" wide. See, for example, the data sheet of the ADC1175.

Have you considered using capacitors across the Optos in order to speed up the pulse? The opto can pass the low frequecny component and the Cap the high freq. You would need circuitry to combine the 2 signals, a type of OR gate.
 
I didnt think of using the opto-isolators + caps, I might have to give that a try, but my isolators are very slow. I'd prefer to use the seperated logic and power grounds on the ir21844 if possible.

I dont think the data sheet says anything about how to handle the 2 seperate grounds. It looks like the power ground is capacitivly coupled to Vcc?

I do have bypass caps all over the breadboard. Is it possible to have too many caps? and do I need to mix it up with several types/values? or can I just use a handful of 10uF's?

Also, I noticed that there is quite a bit of noise even with the motor disconnected. Is that a bypass cap issue then?

thanks for the advice
 
10uF is way too large to filter high frequency noise from ICs. You need 0.1uF or 0.01uF (the smaller the cap the lower the parasitic inductance=more effective high frequency bypassing and the closer to the chip it has to be to be effective). You can use a few sizes progessively larger (starting at 0.1uF or 0.01uF), with the intention of different caps providing different low impedence paths for different frequency noise. But if you get into all the theory mumbo jumbo, this can cause resonance circuits which produces spikes in your impedence which can make the filtering less effective over the frequency range of interest than if you just had one capacitance. The ideal thing to do is to use many small capacitors of identical value (that's what's used for high frequency processors and Ghz chips, etc.).

For all intents and purposes, just use a single (or maybe two) 0.1uF caps or 0.01uF caps as a rule of thumb for ICs. The higher the frequency, the smaller the cap you should use. 0.1uF is regular I guess. 0.01uF is faster. 10uF is too much. I use those to bypass for an analog sensor that runs off the PCB by wires for some distance because the small value caps are ineffective if too far away from the component you want to bypass, but they are more effective for higher frequency noise.
 
Last edited:
Ceramics are good for high frequency noise bypass, and just put it in parallel with a single high value tantalum cap. Electrolytics are good for hold up power but larger values have relativly high internal resistances which make them poor at being bypass caps in the higher values.
 
Yeah.

Ceramics = high frequency, good for bypassing. But can introduce piezo noise so dont use them in amplifiers or filters, etc. You *might* need to use these in high frequency amps/filters, but only if nothing else can do the job (and im pretty sure there are polyester or other caps like that that can do the job better, but come in smaller capacitances than ceramic)

Tantalum- can come in high values, not as good in high frequency as ceramics but still good. introduces no piezo noise so you can use them in amplifiers and stuff well

Electrolytic- what Sceadwian said

There are also different dielectrics for ceramics that have varying capacitances at different temperatures and DC voltages. Check it out. www.avx.com has some good app notes and info if you really want to know.
 
If you're really having noise issues as bad as you say I'd pick two ceramics, a low value and an high value ceramic, and for real power buffering add a generic high value tantalum cap to the mix on your power line. Add mid value ceramics to every line that produces noise, and if possible use an opamp or schmit trigger buffer on any MCU output's that produce PWM as the lower current you draw from an MCU's output pins the better. A heavily or randomly loaded MCU is goign to produce incredible amount of noise. So keep the power load OFF the MCU. An opamp or buffer can be bypassed to almost any desired rise/fall time with the right capacitor.
 
When you say "breadboard", you don't really mean the white plasticy things that don't-have-any-current-rating-limits because even the manufacturer isn't willing to rate them?

High frequency/high power circuits and solderless breadboards just don't mix. Leave the control logic on the breadboard and build up the power circuitry on a dedicated PCBoard, dead bug style, or at the very least on a seperate area of the breadboard.

When the circuit says "seperate grounds", it means two well connected ground circuits that are tied together at a single point in order to elimintate ground loops. They're still electrically connected...
 
I would limit bread boards to the following absolute maximum ratings:

Current: 2A - Providing you don't change any connections whilst current is flowing.
Voltage: 60V - The maximum voltage you can leave uninsulated according to the BS EN safety standards.
Frequency: 10MHz - Sinusoidal.
 
Well I think i have a duct tape solution to the problem, I ran the encoder signals through a couple decade counter ic's set up to divide by 2. This gave a wider pulse, Which could then be digitially filtered in the FPGA. Seems to work pretty good.

Here's a pic of the 'final' circuit:
**broken link removed**
It's a horrible layout and I'm half tempted to redo it, but it appears to work, so why fix it if it ain't broke...

Here's a pic of mosfet failure without current limiting:
**broken link removed**
We were testing the pendulum and the motor quit working all of a sudden, and i looked back at the breadboard to see it blowing smoke.
I since added some current limiting as seen in the first pic, (4 large resistors +ic's).

Some movies of the pendulum in action:
**broken link removed** (3.9mb)
**broken link removed** (1.4 mb)

We still have some control work to do, but I'm quite happy we made it this far.
 
Nice, just looks like the control loop needs to be tweaked.
 
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top