I'm thinking about getting an oscilloscope and was wondering if anyone had some recommendations. Fluke seems to be the big dog, but I can't justify $1200 US (not yet anyway) I'll mainly be using this for my car (testing ecu and sensors) and my pic projects. I'd like to spend under $300 US. Any help would be appreciated. Thanks, -Lee
If you're not after a top-range fancy one, check out ebay for some second hand ones. I got a bargain on my duel trace Beckman, only £50-60 (can't remember exactly). _C
I picked up a Tektronix 453 for $40USD off Ebay. Bought the service manual and 4 hours on a sunday and it works like a charm. Nice simple discrete stuff.
Thanks. Any features I should look for? I know what I want to do, just not what it will require. So it's somewhat tough to decide.
We have used Tektronics, Agilent, and LeCroy at work. While a TDS7404 will be WAYYYYYY out of your price range (lets just say price is always listed as CALL), Tek makes very good machines.
The tektronix 2465 I have works like a charm. But I cannot really reccomend anything else, I haven't used any others however this one can do quite a lot of things, and can sample at a very high rate. The osci I had before this one was a really old one. Vacuum tubes and all. And, it only went to 300khz rate, dim at that, and weighed 100lbs. It's still around here somewhere. It did 4 channels as well.
Tek465 is a good one. 100MHz I beleive. To sum up: You want something that can be used on automotive voltages and also microcontroller/circuit speeds. You probably dont need anything more than 50MHz. Also, the older scopes were built tough (hundreds of discrete components rated way higher than the application they were used for) they can handle messy automotive voltages. Digital scopes are made mostly from ICs. They are harder to troubleshoot and difficult to repair (not to mention they cost more). Documentation is also harder to come by. My TEK453 is a 50MHz scope made in 1963. I bought it off Ebay for $40, and got the service manual for $5. When I got it, the trace was attenuating off the left side of the tube. The service manual gave me 3 steps to verify that the horizontal control was the problem. From there, you look at the blown-up schematic for that part of the scope. The schematic tells you the settings for the scope, and gives you the voltage readings you should expect at each node. After roughly 4 hours (2 of them was simply dismantling and inspecting for obvious burnouts) I was able to get a funky voltage reading across a NPN transistor. I replaced it with your common 2N3904 and the scope works fine now. The whole thing is all resistors, pots, caps, diodes, and transistors. It doesnt take all that much to fix. But be careful, dont try to fix a scope with serious problems like the transformer or the display tube. These parts are able to be sourced, but do require some experience (that i personally dont have) to fix properly. One other note: These older scopes do not have storage capability. This is where digital scopes have an edge. If the signal you want to see is a one-time-shot (not constantly repetitive) then you need a storage scope to see the signal. If you have a $200 bugdet, here's what I would do: Find a TEK465/464 on Ebay that guarantees it works and has been checked (possibly calibrated). $200 is the most you should expect to pay.
Thanks for the help. One of the things I'm looking to do in the near future is decode the signaling bus on my car. Here's a sample scope from the mfg's electrical training workbook (got to love ebay ) Do you think I'm better off with a storage scope? (Perhaps I can just videocapture the scope screen) There's a larger image here with what the legend names.
For that signal, you definately need a storage scope. Standard scopes work by triggering on a repetitive signal. What you see displayed is that signal being repeated many more times than you can determine with your eyes, but since they are repetitive, you see the overlay of many signals being the same. If the signal varies, you'll see a blur (the same effect as a photo taken with slow shutter speed of a fast moving object). Ive personally haven't tried to play with signal equipement like this for hobby use, so maybe someone else could give a good suggestion? Something better than "I use this at work, but the equipment costs thousands".
I have nothing probative to add to this, just the fact that $67,980.00 USD is a shitload of money....wow....http://www.avalontest.com/shop/tektronix_tds7404_av_.htm
There is also the option of using a PC based oscilloscope. Some are parallel port based, others USB, some even are souncard based, so you only need software. This will give you a storage 'scope and if you have an old laptop ( even a 486 ) you may be good to go mobile. Just think, an excuse to get some more hardware! I have used one of the older parallel port scope models on a 486dx2 laptop running Win95, and it was quite serviceable.
Believe it or not, a scope is on my christmas list (near the top too) so i am on the lookout. Its a shame, i have had some really good ones in the past but have been temptes to sell the mfor what i see now as stupid sums of money. I had a solid state storage oscilloscope, tektronics with 100s of buttons, knowbs etc, ex MOD and sold it for about £50 might try and buy local though...
Thanks again for all the pointers and suggestions. I'm thinking that I'm going to hold off for a bit on the scope purchase until I really understand what I'm going to need out of it. Though I'm thinking for this particular problem, I might be able to take a short cut. It certainly seems worth a try as I have all the parts. From what I've read, all of the ECUs commincate in serial (1-7) in every packet frame. It's a pulsed signal so I would think the width of the signal couldn't be modulated as they all have to have their turn in each packet and from the sample scope shots, the packet seems to be the same duration. So I'm thinking that each transmission has to represent a fixed number of bits, with x amound of bits allocated to each ecu. I'm wondering if I can't just use a pic to sample it (not adc, just hi/low) and output the data serially as fast as possible. The limiting factor being the pic clock and the serial connection I would think. I can cause specific events by say, pushing certain switches (such as using the door locks) so I could probably get a good idea if I'm sampling it at an adequate rate by observing events I caused. I'd have think it through a bit, but perhaps I can output serially in a burst between packets to avoid latency issues, though the packets appear to be longer than the time between packets.
PIC, maximum serial transmission rate? Anyone care to guess the maximum serial bit bang transmission rate of a pic clocked at 20mhz? Ignoring the limitations of most PC UARTs (115200). The pic code would do nothing else but transmit the value of 1 pin (2 if I can get the speed high enough). I'd have to craft the assembly carefully to interleave the sampling and transmission timings to try and normalize things. I realize I'd have a dead spot in the sample during the stop and start bits. Do you think 230400 is possible or maybe even 460800 baud? I'm guessing the latter is unattainable. Given 5MIP, that would only leave me 10 instructions to transmit 1 bit. Thanks, -Lee
Ebay is your friend. If you want a storage scope and don't have thousands try looking around for PC scopes, that is very fast a/d converters that plug into your pc and display the trace through some software, most of them support storage. They should only set you back about half the price of a new standard scope.