Are there any special recommendations for prototyping microprocessor circuits on plug-in breadboards?
What is the practical upper limit for the clock frequency? On a few occasions, at 20MHz, the code has not run properly.
What would cause this, and are there any remedies, apart from a completely soldered circuit which I would prefer to avoid in this instance?
I know there are a few proto-board knockers out there. Please don't ruin it for the rest of us.
Ralph Metzner
Those plastic breadboards are awful. Better to use surface-mount adapters on a slab of FR4.
I like this live-bug style because the connections are so obvious. The plastic breadboards, and things with wires on the bottom, are confusing.
You can write comments on the FR4 breadboards, and you can keep them for future reference.
It has been said a number of times, that those plug-in breadboards are crap. 1) inductance along the connection lanes. 2) coupling capacitance between lanes. 3) solder flakes from plugging & unplugging wires/leads drop down causing intermittents and shorts. 4) poor/inadequate bypassing capability due to #1 above as well as long lead length, eg: simple 16 pin DIP "bypass" lead length cannot be shorter than the length of the DIP. 5) *NO* ground plane possible to help mitigate #1.
Might be semi-useful for microsecond (rise, fall) signals but not nanosecond signals.
Bottom line: special recommendations for prototyping microprocessor circuits on plug-in breadboards is: DO NOT USE THEM!
Now you can avoid soldering, by using MillMax contacts soldered into a PCB, or sockets (almost equivalent but a bit more capacitance and inductance and so more frequency limiting).
Are those little rectangular, labelled proto boards your design? They look SO neat and useful!
Is this a microcontroller with built-in memory, or is this some ancient chip with external ROM? If it's a newish micro, your best bet is to make sure that the bypass cap has as short a run as possible from ground to power -- and if there's more than one power and ground lead, that each one is bypassed, and has a good connection to its respective bus.
I like protoboards myself, but there are things they're good for and things they're not.
-- Tim Wescott Wescott Design Services http://www.wescottdesign.com
keep the microprocessor, it's powersupply it's clock/crystal off those boards, only use solderless proto boards for slow peripherals.
I wouldn't trust them above 500Khz
capacitance beween tracks, contact resistance etc.
Solder the fast part of the circuit, and run wires to the slow part on the breadboard. or use a pre-built circuit like "arduino", "beaglebone", or "msp430 launchpad" for the fast part.
Else run the circuit with a slower clock on the breadboard,
-- umop apisdn --- news://freenews.netfront.net/ - complaints: news@netfront.net ---
On a sunny day (Tue, 26 Aug 2014 12:49:36 +1000) it happened Ralph Metzner wrote in :
Right, those wire plugin boards, if that is what you refer to, apart from the unreliable contacts, will cause big problems near or above 20 MHz if you do not know exactly what you are doing.
Just get a soldering iron, and if you MUST use wires, this goes to 18 MHz:
You can go to some GHz too when using SMDs:
Anyways that is how *I* prototype.
The main point is that you understand what happens in such a piece of wire. As long as its length does not come close to a significant part of the wavelength you do not have to worry too much, Inductance should stay low too. And I am sure not every signal on your board is 20 MHz, so use brain. Decoupling is an other requirement that I do not see happen with them plug-in wire boards, as you can see (or not because the SMDs are so small) from the pictures, decoupling caps are sometimes needed right across the IC pins.
Those 'veroboards' with isolated isles are way better, and can be kept in use once it works.
Have you really attempted to run something like 20 MHz on those boards?
Building short wave (3-30 MHz) radio receivers with non PCB techniques, the performance depended greatly due to the physical layout.
TTL minicomputers in the 1970's contained a lot of wire wrap backplanes. However, these were running at about 1 MHz system clocks.
I definitely not would get any microprocessor address or data busses into a protoboard or even a 0.1" Veroboard, but of course the microcontroller I/O pins would be OK for such boards.
Makes me want to reach for the spagetti sauce ;-)
Ralph Metzner
Are there any online tutorials for this type of construction?
I am very interested in alternative prototyping methods, apart from the "rats nest" ones and the usual veroboard.
Does anyone have any others they can suggest?
Ralph Metzner
There is a Vector board with a full GND and Vcc plane, other than that I'd go with a real spin of the board.
Cheers
Typical of wiring; but more on this below.
Typically 2pF, which as long as you're aware of it in capacitance-critical apps, is usually not too bad. And can even be a little useful.
I'm sorry you take such poor care of your boards. ;-D
Also, bent springs. $%#@!
Worse than PCB obviously, but not a problem with DIP. In Ye Olden Dayes, they did that all the time -- cap from corner to corner beside the chip, no ground planes (and lots of EMI!).
Not necessarily. I find much improved results from using the kind mounted on a baseplate. Even just grounding one corner (scrape off the paint and use a screw or clip or something) to the circuit helps.
A trite example, but I've done avalanche pulse generators before. Seems like you lose a nanosecond or two in the board, though. Not sure how lossy the (acetal?) base is at frequencies. The foam tape (if it's adhered to a base) probably isn't much trouble (it's foam!). Trite because, you get what you get from the avalanche to begin with; wherever that signal goes, it remains that signal, give or take losses and dispersion and such.
The most aggressive freq * power circuit I've done is a 500kHz resonant converter at 20W or so; amazingly enough, 24AWG jumpers and spring contacts don't much appreciate 5A at that frequency. :)
As long as the contact resistance doesn't eat into your circuit's performance, those boards can be used for quite a bit above ratings: I wouldn't suggest putting 1000V in adjacent slots, but I've never had 500V cause breakdown, and for higher voltages you can simply skip a few over. Above 5A seems troublesome (the most I've done is around 10A, which I don't think burned the contact).
-=-=-
I wouldn't normally suggest circuits over 10MHz bandwidth; you can try building a VHF radio, or ECL logic, stuff like that, but you're more likely to get a crummy oscillator instead.
The main thing that you will get, from breadboarding stuff like microcontrollers, is signal quality issues.
MCU pin drivers are usually 50 ohms or lower Rds(on), and the 'traces' going between points will be quite high impedance. Even if you route the wires close to power supply buses, their characteristic impedance won't be any less than 100 ohms. Typical "through the air" routes will probably be in the 200-300 ohm range.
The loads will be higher capacitance because of slot-to-slot plus pin capacitance, which will resonate with the effective inductance of the wire.
Simple solution? Source terminate the MCU, using a ferrite bead or series
100-220 ohm resistor. This will be most important on fast changing, high data rate signals, like SPI.Mind which signals are level triggered and which are edge triggered. Edge triggered signals *will* bang repeatedly from bouncy edges. An SPI input will read trash. Filter that first and foremost. Level signals (like data lines) aren't as critical, but do mind crosstalk. Port pins aren't usually doing fast things, but if you're bit-banging shift registers for example, that's another thing. Toggling whole ports at a time will probably cause global crosstalk issues (ground/supply bounce, that sort of thing), so avoid that too.
Crosstalk will be significant, especially if you run a bunch of active signals together through the air as a cable. You can run extra ground-to-ground wires between points of use to give some ground reference to the cable, and you can add some C or R+C to ground at point-of-use on active and/or inactive pins to shunt or dampen the edge ringing.
There are still some things you won't be able to do, like getting clean, sharp signals right off the MCU. Those you'll have to buffer at point-of-use. But you should buffer MCU signals anyway, before they reach the outside world. Good safe engineering practice. A lot of the things, like ferrite beads and R+C damping, won't be relevant on a proper PCB model of the same circuit, but signal quality issues never just disappear, and treating those problems on the breadboard will give you an exaggerated idea of what to expect on a PCB. The development time saved in breadboarding is well worth it, so don't believe the negative nancies that would have you believe otherwise. :-)
Tim
-- Seven Transistor Labs Electrical Engineering Consultation Website: http://seventransistorlabs.com
No, they are from Bellin. Mouser and Amazon sell them.
The other trick is to cut up the copper with a Deremel and a carbide dental burr; I get them on ebay.
I don't know of any. Just practise a little.
Use Bellin soic adapters.
Attach them to the copperclad with foam tape.
Attach copper sub-islands with krazy glue.
SMB connectors are great for fast i/o
Use gold plated FR4!
Or maybe standard DIP sockets with the legs bent out to the side. The plastic body could be fixed to the board with industrial quality tape.
I might also try isolating sections of the board with a small hole saw, just deep enough to remove the copper. Has anyone ever seen a square one?
Hmm .. amd much cheaper from China on eBay.
Which raises the question, how do you keep the copper boards free of fingerprints and tarnishing? Spray it with solderable PCB laquer? Rubber gloves?
Do you need to keep adjusting your iron temperature? That copper would draw a lot of heat compared to the SMD's.
What are the two rows of solder blobs for on this board? There seems to be nothing connected.
Thanks for your advice.
Ralph Metzner
Until people got tired of their designs not working and the chip makers started putting Vcc and Gnd in the center of the package. Now most people use better PCB designs and chip packages are much smaller often with multiple power and ground pins.
-- Rick
Sure. Don't do it! Anything faster or larger than an (original) 8051 will cause all sorts of grief. Use an eval board or one of the
*cheap* processor boards.
Not surprising. I've gone much faster (~50MHz) than that with wire wrap but took special precautions to do it.
Lead inductance, primarily on the power rails but elsewhere as well.
If you enjoy wasting your time, go fer it.
I'm surprised JL doesn't use SSMB - supposed to be good to 12GHz instead of just 4GHz like SMB. Still, thanks for the pointer, I've been using SMAs and the need to screw them on can become a pain.
I tin whole boards quickly using a bit of used-up (saturated) de-soldering braid. Use a wide tip and you can do it quickly. The braid's residual flux means it works well.
I make my own prototype boards using toner transfer, single-sided (leave the back plane intact by covering it with tape). I can go from on-screen layout to a tinned board in 30-40 minutes.
Clifford Heath
Although even back in the Z80 days, that wasn't uncommon -- probably a practicality problem for a huge DIP40 or what have you, even back then. Your average ATmega32x thingy has them in pairs, in the middle of both sides -- assuming you get the DIP40 for breadboarding at all.
Otherwise, QFP breakouts are cheap and easy, and the kind with pin headers are even better for breadboarding -- only the programmy bits get to the board, who needs to worry about supplies?
Tim
-- Seven Transistor Labs Electrical Engineering Consultation Website: http://seventransistorlabs.com
Or just solder the DIPS in. Bend any non-groundable pins horizontal and trim maybe. Solder as many pins to the ground plane as possible.
Leaded ceramic caps make nice combo bypasses/standoffs/places to clip power supply leads. A megohm resistor can be a good standoff, too.
Old TV-type terminal strips are handy, too.
Use gold plated FR4! I had a board house plate up a couple of square feet of double-side copperclad. It solders beautifully and doesn't tarnish.
Kapton tape can be fun, too.
I have a Metcal iron, which pours out heat as needed.
Wire vias that connect the top and bottom ground planes.
-- John Larkin Highland Technology, Inc jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.