I accidentally picked the wrong MJN relay. They come in a flange mount form and a non-flange mount form. The latter is what can plug into a normal plastic socket. Luckily, I stumbled into noticing this before placing an order.

The non-flange mount part number is MJN3C-IN-AC120. Mouser doesn't stock them, so I found them at onlinecomponents.com through Google.

MJN3C-IN-AC120 3PDT Relay @ OnlineComponents

PTF11PC MJN Plug-In Relay Socket @ OnlineComponents

 

Just a quick update. The case and other parts arrived last Friday. CAD drawings were put together and the panels were run on the mill to make all the necessary parts and brackets mounting holes. Once the case was re-assembled and the relays, BAP1950 and USB-6211 were laid out, we realized we were going to also need a precise collection of fasteners. I initially thought we would be able to get away with scrounging in the parts bin. McMaster-Carr delivered the necessary screws and nuts this week, and everything is now bolted together.

I sent an email to Les @ Applied Power Systems, this time asking whether or not the MJN relay inline with the 3Phase Ref wires to the BAP1950 was necessary. I supplied the separate Modified Sicon Controller Schematic and "NewDynoControlBoard" Schematic. Les asked for a complete schematic on one piece of paper. I had been meaning to do this, but this was the kick-start to get it done.

Three-Phase Eddy Current Dyno Circuit with Dual SCR Controllers (pdf)

There's a lot of assembly work to do now, and a little bit of CAD work left. Then it's software design and testing.

 

The details will always eat up more time than you can imagine. Hope the critter works right the first time.

You might want to add terminal labels to all your control lines for assembly and later reference. I've worked with all white wires and no labels - gets a bit tedious verifying all start and end points. Lots of emergency masking tape labels.

 

Thanks for the warning. The potential headache of tracing thirty white wires in close quarters started to weigh on me. I wanted a shrink wrap label printer, but my first search turned up options beginning at $1,000 USD. I took another look on Google this morning and found the K-Sun LabelShop 2012XLST Shrink Label and Wrap Label thermal printer for $400 MSRP. I spoke with Bill @ WireMarkerPlus.com through a nifty online chat window. It's more expensive than writing on tape under clear shrink wrap, but it should get the job done.

I realized on Friday night that the MK relays will not work for switching the six SCR Gate and Cathode signals. The cathode signals are pushing 380VAC and the MK relays are rated to only 250V. This is the reason we spec'ed a different relay, the MJN series 3PDT piece, on the 3Phase Ref connections. The good news is that the same MJN series relay, the MJN3CIN-AC120, can be used for all three relay requirements.

This weekend I aimed to begin bench testing the DAQ side of the new controller. The two input data signals are the variable reluctor (VR) waveform representing engine/dyno RPM and the 3mV/V output signal from the load cell. The VR signal is being passed into a LM1815 circuit in Jean Belanger's Dual VR Conditioner Board v1.1. The Dual VR Board outputs a digital pulse train that I previously planned to enter into one of the NI USB-6211's Digital Input channels. After reading sections of the USB-6211 User Manual, I found out I need to connect it to a Counter/Timer type digital input channel. The unconditioned load cell output signal is connected to an Omega OM5 Full Bridge Strain Gage signal conditioner module. The scaled 0-5VDC output signal is connected to an analog input on the USB-6211.

Short of using an actual VR sensor held closely to a trigger wheel being spun by an electric drill, the MegaSquirt Stimulator v2.0 was the only immediate option for bench-testing the DAQ connections. It uses an IC1 555 pulse timer and a few other pieces to generate a VR-like waveform. Jean helped set up the Dual VR Board, and modify the Stimulator VR output signal, over email on Sunday night. I put together an "instant VI" LabVIEW application that read the frequency of the Dual VR Board's generated pulse train. Yesterday, we were able to vary the RPM reading in LabVIEW my adjusting the RPM control knobs on the Stimulator. This theoretically means that the RPM data input functionality is ready for real, off-the-bench testing. Realistically, in the next week, more time will be spent developing the LabVIEW VI and bench testing.

 

Probably a good catch on the relay contact ratings. It may make no practical difference, as those lines should not be pulling current so there won't be arcs to break. But it's always a good idea to have proper ratings on everything.

Yes, for RPM, you want to count pulses per unit of time.

Long ago, I worked with a couple other guys to locate a bad wire in a computer. Many thousands of identical white insulated wires running here and there. We used a broom handle, a screwdriver handle, and finally a crochet hook to locate the bad wire. It took three hours to make the fix. Gotta have labels.

 

Long ago, I worked with a couple other guys to locate a bad wire in a computer. Many thousands of identical white insulated wires running here and there. We used a broom handle, a screwdriver handle, and finally a crochet hook to locate the bad wire. It took three hours to make the fix. Gotta have labels.

I would love to hear this story. Any computer repair that involves a broom handle, screwdriver and crochet hook rates at the top of the charts in my book.

 

Actually, it was pretty simple. Changing out the logic at either end did not produce results. This was a Univac 1206 (Navy designated it a CP-642A). There were 13 chassis inside the main case. The chassis were about 3' by 4'. Each had thousands of edge connectors with something like 20 contacts. A 2" by 3" printed circuit board holding one to three logic functions was in each edge connector.

Each chassis had a row of 8 connectors along the right and left edge for wiring runs to other chassis and the maintenance panel - I think you can find an illo of one of these online. Yeah, here's a link - http://ed-thelen.org/comp-hist/univac-ntds.html.

Anyway, the side connectors were inserted and disconnected with a rack and cam setup to push in and pull out the mating connectors. Everyone carried around a 3/8" driver with a 5/8" socket to undo door latches and to run the chassis connectors in and out.

We had a failure that was intermittent, and was not affected by the end point logic boards. But you could provoke the halt by shaking the computer. This was fun - the beast weighed in a 2200 lbs. But it was mounted on sprung shock absorber mounts.

Finally, we determined that kicking the left side always brought on the fault. So we pulled the left side panel to expose the wiring. Judicious whacking with a broom handle localized the fault to a vertical range of about one foot. Using a screwdriver handle, the area shrank to a couple of square inches. Fingers eliminated some further wiring, and we got a crochet hook to pull individual wires to locate the one with the problem.

The break, or intermittent discontinuity, was away from the side connectors, so we just cut it back toward the terminations and spliced in a new wire. Just another fun problem troubleshot successfully.

That wasn't even the weirdest fix.

 

Finally, we determined that kicking the left side always brought on the fault. So we pulled the left side panel to expose the wiring. Judicious whacking with a broom handle localized the fault to a vertical range of about one foot. Using a screwdriver handle, the area shrank to a couple of square inches. Fingers eliminated some further wiring, and we got a crochet hook to pull individual wires to locate the one with the problem.

Hahaha that is great! I laughed out loud when I read the bold'ed part above. I'm going to have to figure out a way to carry a broom handle in the toolbox. I can't even begin to imagine how many wires you are talking about, I got lost in the listing of components and sub-components. The 5/8" socket and cam-driven connectors sound really neat though.

I realized on Saturday that both the VR and strain gage signal conditioners require +5VDC, and that I had not planned for a 5VDC supply. The Omega OM5 full bridge strain gage needs 200mA @ 5VDC. The half of the Dual VR Conditioner Board v1.1 that I am using needs ~5mA @ 5VDC.

I took a look on Mouser for regulated and caged (enclosure) 5V power supplies. The company Mean Well offers a 25W 5V 5A regulated DC power supply for $34 with a relatively small footprint (3.1" x 2.0" x 1.1"), part number RS25-5. Datasheet

If I wanted to use a supply like the RS25-5 with extra capacity for additional 5V load in the future, is there anything I have to worry about if I only hook up 205mA of load to it? I suppose I could add fuses, but I am not familiar with fuses outside of 5A-30A automotive cartridge fuses. Would I want to install a separate fuse for every device pulling 5V power from it?

 

You would want to use automotive fuses. They are rated to blow clear at low voltages - something that 250 VAC rated fuses are not. The little supply sounds like it might be a good move. The 5 amp capability means lots of 200 ma loads can be added in the future. The price is pretty good.

As to a separate fuse per load - that is probably going to be awfully busy. See if you can find some value like 1/2 amp, and fuse branches based on that. Just cluster loads so the draw is less than the fusing. Automotive fuses may be hard to find at values less than 5 amps, though. The number of critical 5 volt circuits running unfused is just about equal to the total number of critical 5 volt circuits. We always assumed that a transistor would fail in such a way as to protect the fuse.

If your new control panel has a sense input for it, you might arrange for it to monitor +5 out at an important sensor so it can shut down the dyno if something goes dark.

Note as to fault-localizing technique above - if you can't do something in a way that would worry the people nominally in charge, why bother fixing problems?

 

I placed the order for the RS-25-5, a couple of 1Amp ATO size automotive blade fuses, and two inline wire-terminated fuse holders. Fuse holders with bolt holes or other mounting provisions would be ideal, but there didn't seem to be any available for automotive style fuses. Not a big problem.

1A ATO Automotive Blade Fuse

ATO Size, Wire-terminated Blade Fuse Holder

Mean Well 25W 5VDC 5A Switching Power Supply

One part of the new controller that I have not begun to think about yet is the 120VAC wiring inside the aluminum case that houses all of these parts. I have seen three individual ~16awg, stranded core wires loosely run inside boxes to care 120VAC and I have seen home construction type Romex cable used. Any recommendations? I'll try and post a drawing of the 120VAC wire routing that's needed.

 

The 16 ga off a terminal strip is going to work just fine. Use standard colors and just route neatly.

 

Just a quick picture before a full update.

 

Very sanitary layout and wiring. Those Amphenol military connectors are the best (expensive critters, though).

It's so neat, it just has to work.

 

The hardware has not been touched much recently. A 5VDC supply was added for the variable reluctor and strain gauge signal conditioners, and two switches were added to the front panel. This is largely good news though because it reflects the fact that the hardware has been working nicely.

Progress is being made on the software side, but like all projects, the details are really showing their teeth. We did have a chance for a very basic, preliminary test last week to use a closed loop PID controller to regulate the dyno's brake and bring the operating RPM of a diesel down from ~970RPM to ~910 RPM.

3/21/09 Engine RPM versus Control Output Signal Chart

3/21/09 LabVIEW Screenshot

I'll post a few pics once the new switches are wired in.

The software development thread is over at the National Instruments discussion board. Here's a a link.

 

The details always eat you alive - but then the hardware is working correctly. Should be interesting to be able to program up a full test cycle.