Monday, November 12, 2012

A Silly USB Stick, It Presses 'F1'

So it happened to me. Recently I received some alert that my colocated servers were down at a local datacenter. There had been a power event where the UPS units there had failed causing a reboot of everything in this specific DC. While driving the the DC at 1am to figure out what was going on, I received alerts that some of my servers had came back up, while one had not. Upon arriving there were already a ton of people there dealing with their own equipment. Looking at my specific servers I could see there was a bad DIMM error on the front display, classically flashing orange. "All I need to do is press F1 To Continue" I thought, but unfortunately all of the crash carts were in use by other people there. Anyone who has Dell servers has seen this message during boot at some point if there has ever been any issues with their server. It is a classic error message which I wish Dell had always by default just made the server boot if possible. Because of the queue of people waiting for crash carts, I had to wait until one finally became available to press 'F1' to get my server alive. This could have easily been avoided if I had set this specific server to boot up regardless of error in the bios, but it was too late for this now.

The solution? Well, make sure your servers are set to boot regardless of error in the bios. But a backup solution is my "F1 USB Stick":

What does it do? It is a UDB HID Keyboard that presses F1 repeatedly! That's all!

Simply plug it into a server and it presses F1 for you. No keyboard needed! Your server will be back up in no time.


I had fun with this one, looking at the kernel messages upon plugging in and you will see this:

usb 3-2: new full speed USB device using uhci_hcd and address 3
usb 3-2: configuration #1 chosen from 1 choice
input: Brad Boegler  Press F1! as /class/input/input3
input,hidraw0: USB HID v1.11 Keyboard [Brad Boegler  Press F1!] on usb-0000:00:1d.1-2

It is based on a Microchip PIC18F14K50 USB enabled microcontroller, a 12Mhz crystal, and the necessary passive components for operation and USB communication. I added a few LEDs in there just to be able to see what was going on. I wrote the code in Microchips C18 which utilizing Microchips USB libraries made this as simple as can be.

I will be keeping one of these in my laptop bag at all times... just in case. ;)

Monday, October 8, 2012

Hydrogen Line 1.42Ghz RF Front End - Radio Astronomy

After finally completing the power supply for my RF front end last week, I spent some time tonight assembling the power supply onto the back of the feedhorn assembly and wiring all of the amplifiers and downconverter to the necessary voltage supplies. Here is a look of what it currently looks like:

From the front, you have the feedhorn, LNA, 1420Mhz cavity filter, downconverter, low pass filter, second LNA, then the power supply on the very back. Now that it is completed, I performed some testing tonight to check operation, sensitivity, and power supply stability. So far everything is looking good, the video below shows its operation on my bench for testing purposes:

Saturday, September 29, 2012

Radio Telescope Power Supply - Progress

Today I received my power supply boards for the RF frond end of my radio telescope. Due to the number of amplifiers and the voltage requirements of my downconverter, I needed a supply that provided a stable and clean +5V, +10V, +12V, +15V, and adjustable 8V to 15V output.

Current requirements on all outputs is very small (less then 20mA at each respected output) with exception of the 12V supply which powers the mini-circuits ZRL-2400LN low noise amplifier. It has a 350mA @ 12V requirement so I have utilized the aluminum enclosure to sink some heat off of the 12V regulator. I slightly miscalculated the boards size for the aluminum enclosure I had chosen to use for it. Due to the mounting standoffs inside the case, I was not able to mount the board flush against the side of the enclosure... although it was nothing a band saw couldn't fix. All voltage outputs are passed through the enclosure using feed through capacitors.

There is nothing really exciting about this supply other than the fact that it is now completed. Power supplies are probably the least exciting components of any project to build, even though their importance and stability are critical. Now that it is completed I can install it on the back of my RF frontend and start prepping the actual mount for the dish.

Here is the final assembled board mounted in its shielded enclosure:

Thursday, September 6, 2012

A Repair Story: EIP 545A 18Ghz Frequency Counter

This post is about a frequency counter, specifically an EIP 545A 18Ghz frequency counter that I had purchased off of eBay. The purpose of this post is to describe the steps that I took to repair this specific counter as upon receiving, it was not working. I consider myself very good at troubleshooting and here I will outline the steps I took to get his unit repaired. The physical repair itself was simple as it ended up only being a single cap that was bad, although the steps taken to isolate and troubleshoot this unit to locate this specific cap were very complex. As I repaired this unit I could see remnants of previous people attempting to repair this unit, who had ultimately given up and failed. While normally I would never write about specific repair steps that I had taken, this one was specifically fun, fulfilling, and worth talking about. Hopefully the troubleshooting procedure and methodology I describe here can be of use to someone else.

This specific EIP frequency counter I purchased is older, from the 1980s but is a very solid unit and when combined with a stable 10Mhz frequency reference it is extremely accurate even for its age. The range of prices for these specific units is all over the map, some guaranteed dead units going for as little as $75, with fully functioning and calibrated versions with all options up into the thousands of dollars. Most working units are in the $200 - $500 range in various conditions. This is a steal for a counter of this range, resolution, and accuracy.

I went with my staple bid on the unit that was listed by a seller who did not deal with test and measurement equipment. It was listed as working, but 'we don't know anything about this unit so we are selling it as-is' by the seller. The auction showed a picture of the actual unit powered up with all zeros on the display and no reported errors during power up. The seller also stated if it didn't work it could be returned as well. I purchase a lot of equipment this way. There is a risk involved as it could have been messed with and someone is lying and unloading it, but often I end up with a great piece of gear in these auctions. Based on all the info I had I went for the purchase.


Upon receiving the unit the first thing I noticed is the calibration stickers had been broken indicating someone had been inside the unit, big red flag. Why was someone inside it if it was working?  I wanted to know so I took the main cover off and looked inside. Other than being very dusty which I promptly blew out with an air compressor, nothing seemed loose or broken from shipment. Some board clips were broken indicating someone had removed a few modules at one time, the plastic gets brittle with age. There were a few boards missing but these were to a few options that this specific unit did not have installed that I was already aware of so we should be good. The good news is it looks like no modules were cannibalized from it for a different unit. To the bench to power up:

EIP 545A Frequency Counter

Upon power on the display came up, registered all zeros. So far so good. The top of the unit had a couple of the basic self tests labeled right on it, starting with test 01:

Test 01 injects the internal 10Mhz time base through a few multipliers into the counter to verify the time base reference, signal chain, and local oscillator. Unfortunately instead of displaying 200Mhz, the display was all blank.


Now I know why the calibration stickers were broken, someone had most likely been poking around inside in an attempt to fix. This could end very badly.

The good news is all of the other self tests passed. There were no ram / rom errors, all display and keypad functions worked, and overall the unit seemed fine with no additional reported errors. Using my HP 8656B RF frequency generator as an RF source on the counters inputs resulted in the same result, no frequency displayed. So the troubleshooting begins.

Once opened back up, the first steps are a simple visual inspection. Is everything plugged in? Do all cables seem to be routed appropriately, as in not plugged into the wrong locations? Is anything look loose or moving around? Have any zip-ties been cut indicating wiring harnesses have been moved around or replaced? In this case the answers were no. For the most part everything seemed tight and undisturbed.

The basic internal layout of the unit is pretty straight forward. Starting on the left you have a linear power supply followed by a series of removable cards. The digital section is on the left followed by the analog section on the right. On the very right you have the 10Mhz oven reference oscillator and the RF converter / mixer assembly.

The 10Mhz reference oven oscillator is a prime candidate for failure or issues. I had noticed that the main processor board had its own crystal so it was most likely not using the reference 10Mhz for clock. If the reference oscillator had failed it would be an easy explanation for why the RF frontend was not able to downconvert any input frequencies and not report the internal 200Mhz test. I probed the 10Mhz oscillator, and it was indeed oscillating at 10Mhz. I then tried my external 10Mhz source to see if there was any difference, which there was not.

To further troubleshoot I first re-seated all of the modules and connectors I could see. After 30 years connections can become corroded, dust can work its way into connections, and connectors can be come loose from shipping it around. Note at this point I removed the very leftmost card out of the unit. It is the GPIB interface card and is not necessary for proper operation of the unit. Removing it now just removes one additional variable out of the counter. Once everything else was re-seated, I powered up again but still no success, test 01 fails and any external RF input results in all zeros on the display.

The next step was a simple visual inspection of every board. I pulled everything out and looked at each board for any obvious signs of burnt or charred components, leaky caps, chips with holes blown in the top of them, etc. Everything looked clean and pristine. So time to dig deeper.

One of the very best parts of older equipment is the easy availability of service manuals. After a minute of searching I had the complete user / operation manual and the complete service manual. The really cool part of the service manual, and something you never see anymore is a complete schematic for the counter.


With the schematic I now have details on the power supply including correct voltages and test points. Most service manuals, this one included, will also have some simple flowcharts to help diagnose and troubleshoot any operating abnormalities. In this case I found the flowchart for a failed test 01, following this flowchart I ended up with a result of:  'Impedance converter or signal select on A109 bad'. Well, this is good and bad. It's good because at least it is isolated to a specific board. It's bad because this board is a very complex analog board so troubleshooting it will not be fun. Time to check some voltages.

Probing the voltages, everything seemed present, +5v, -5v, +12v, +18v... but -12v seemed to be missing! Before getting too excited, I powered the counter down, unplugged the power cable from the power supply to the main board, powered back up and measured again. -12v was now present. So something in the unit was sinking -12v to ground. Module A109 which had been previously identified via the service manual flowchart was the first board I was going to look at. I powered the unit back down, reconnected the supply cable, then pulled module A109. When powered back up -12v was still present, so this board was indeed the culprit.

Now going to the schematic for this module I isolated the -12v and ground pins on the edge card and measuring for continuity it was a dead short. So something on this specific board had most likely failed closed causing the problems. Going to the schematic I located every -12v entry point to the circuitry, there was a lot. Over 30 in total.

Semiconductors are usually my first guess for failing, so I began to check all of them first. There was only a handful and all of them appeared to unfortunately check out correctly so time to move on to the passive components. After probing a few caps and resistors that were good candidates based on their location (  between ground and -12v ) I needed to segment the board up further to better isolate portions of the circuit. To do this I would make small cuts in the -12v traces on the bottom of the board probing in sections.

By divide and conquering I eventually ended up at a small .01uF cap, C48 to be specific. Removing one leg of this cap, resoldering my cut traces, and reinserting the board followed by powering up and test 01 resulted in a nice clean 200Mhz frequency on the display.


So a single .01uF cap was all of the trouble. A ten cent component was the reason that this unit was removed from service and sold on the surplus market for next to nothing. It's sad in a way that such an amazing piece of gear can be considered obsolete with no one willing to spend the time to fix it. Granted to most corporations this piece of gear was wrote off in the books decades ago and to pay someone to spend time repairing it would never probably happen. But for me... it's a huge win. I replaced the bad cap with a new one along with a few other on the board just because it was easy and now have a rock solid, 18Ghz frequency counter with great calibration on my RF bench for about two hours of time spent troubleshooting.


Saturday, August 11, 2012

New 10Ghz Oscillator

I have been needing a stable 10Ghz oscillator for a few projects I have been working on and was able to pick this brick up for pretty cheap. This is a Digital Microwave Corp 10.210763 GHz phase locked oscillator. This unit consists of a dielectric resonator oscillator that is phase locked to the x100 harmonic of a 102.10763 MHz crystal.

10Ghz PLL oscillator

While not a YIG, it should be more than stable enough for my needs now (I hope). The new 10Ghz project will be coming after I finish up my 1420Mhz front end for the radio telescope which is nearing completion :)

Friday, July 6, 2012

Radio Astronomy Hydrogen Line 1420Mhz RF Front End Testing

I have been building the RF front end for my radio telescope for a few weeks now and am finally at a point where I can sit down and test it. Right now I have the feedhorn assembly completed and the necessary front end electronics mounted to the back of it.

The feedhorn feeds directly into a mini-circuits ZRL-2400LN low noise amplifier. The output of this then feeds into a cavity filter with a center frequency of 1420Mhz. This was a rare but exciting eBay find hat I came across a few years ago, it actually had come off of the VLA in New Mexico. It had been originally tuned to 1430Mhz but with the assistance of a VNA I was able to tune it down to 1420Mhz. This then outputs to my own custom built downconverter which I have mounted in a custom copper enclosure. I will eventually mill a aluminum enclosure for it but this will suffice for now. Finally after the amplified output stage on my downconverter, I pass the IF through a mini-circuits SLP-450 low-pass filter to remove the original source frequencies, LO, and image frequencies. For testing I have my LO set to 1200Mhz which will result in a 220Mhz IF based on the 1420Mhz source. Ideally I will probably downconvert to 70Mhz as there is plenty of 70Mhz detection gear available on the surplus market. I'm still most likely going to build my own I/Q demodulator for the detection side.

Here is a quick video demonstrating the testing setup:

Sunday, June 17, 2012

21cm Hydrogen Line Feedhorn Assembly for Radio Astronomy

I am excited to finally be able to say that my radio telescope is starting to come together. This morning I worked on another major component of my receiving system, a 1.420Ghz hydrogen line feedhorn. This has been a project that I have been wanting to build for roughly 16 years and now that things are moving along I am hoping to have a system together ready for testing within a month or so.

Now for the feed, I decided to go with a rectangular design for my feed instead of a circular one for simplicity of assembly. I went back and forth many times on which design I should use but ultimately ended up deciding on the rectangular feed for several reasons. It is easier to assemble (90 degree cuts are easy to mill), it's based on a standard size, and the material was cheap. Rectangular waveguides are polarized, although for radio astronomy purposes this should not matter as any natural occurring signals would in theory have random polarization. Here is the final assembled version:

While not really a feedhorn as of yet (I have not built the horn) it is a nice waveguide to coax adapter that will be used as a feed at the focal point of my dish. As for the horn, I will have to check to see if I will have any benefit of using one. A horn can provide additional gain from the dish, but it also blocks off surface area of the dish in its shadow. A choke ring on a circular feed would have the same effect in blocking the signal, this is just something I need to research more.

The feed itself is assembled out of 1/4" 6061 aluminum stock that I cut and milled down to size. The dimensions of the opening are 6.5" x 3.25" which is the exact spec of the industry standard WR-650 waveguide which is designed for frequencies between 1.12Ghz and 1.70Ghz. The Hydrogen line of 1.420Ghz fits nearly perfect between these two limits which makes this specific size ideal for radio astronomy. I drilled and tapped 22 holes which have stainless steel hex head screws holding it together. I was very pleased with the final assembly as it has a nice tight fit.

The probe consists of a 4mm section of copper wire which is exactly 1/4 wavelength of 1.420Ghz long and positioned 1/4 wavelength from the back of the feed. The probe terminates to an SMA connector mounted to the top of the feed. I had to mill a small slot into the top of the feed to allow the bottom section of the SMA panel mount jack to lie flush with the inside of the feed.

One note on the WR-650 standard itself. There are commercial feeds available as it is a standard waveguide size, but the cost is extremely high since this this specific size of waveguide does not show up in the surplus market very often. Smaller waveguide standards for higher frequencies like WR-90, WR-42, etc, do show up but it has been extremely hard to find anything WR-650 available for cheap. My total cost to build this feed is about $50.

I have already tested this with my HP 8614A signal generator set at 1.420Ghz and have verified it does indeed work very well. Next steps are to add the mounting brackets to it which will allow me to mount it at the focal point of my dish and also add the additional RF amps, filters, and my downconverter to the back section of the feed. I still also will need to calculate total system gain and noise once completed.

Monday, June 11, 2012

Altera EPM3256A CPLD Breakout

A few months ago I received my newly designed breakout boards from Laen's OSH Park PCB service and finally had time to assemble my new boards. Stepping up from my original EPM3032A board I went to the EPM3256A here in its final assembled form:

With this new board I only broke out about 100 or so pins of the 208 pin package, this will be more than enough for the projects I have planned with it. I added a few features to it including an on-board socketed crystal with options for 3.3v or 5v operation. I also have options to route the clock to the global clock input or a specified general purpose io pin depending on the application.

A couple nice features about the MAX3000 series specifically to the EPM3256A:

- 3.3v and 5v compatibility. Makes interfacing existing logic very easy.
- 256 macrocells available. Small compared to modern FPGAs, but is enough to be very useful.
- 158 Usable i/o pins.

I will be posting some projects based on this board very soon.

New Spectrum Analyzer: Anritsu MT8801C

The Spectrum Analyzer is by far the most important piece of gear for any RF design work, unfortunately spectrum analyzers are also one of the most costly pieces of gear you can buy (A VNA is right up there too, but that's a different post). I have had access to spectrum analyzers at a few previous jobs which is great whenever you need to test your latest RF design. The problem is when you are at home working at your own bench at 2am,  it's annoying to not be able to have access to this gear all of the time. Buying an analyzer is ideal, but costly. Anything new is pretty much out of the question, so the usual source of eBay is the place to go. Both Tektronix and HP/Agilent have some amazing pieces of gear for an 'affordable' amount (the Tek 49N series and HP 85NN series come to mind), the problem is any of these models can easily cost over $1000 in good working condition. The other issue with this equipment (like all older gear) is their age. Since most were used in a production or lab environment, they have been powered on for 8 hours a day for years. This can result in some crt burn in, the devices being way out of calibration, instabilities and other problems as most of the equipment in this class is 15+ years old (note that makes it affordable). I have had a good run with all my HP / Tek gear as this equipment is really built extremely well. As an example, my HP8614A signal generator was built in the 1960s and still works perfect today. So what happens if you want a spectrum analyzer but don't want to spend $1000+? As I found, there are a few options:

1. Buy a really old analyzer. Some of the older HP models will go for under $500. Keep in mind that these models usually have a max frequency range of no more than a few hundred Mhz.

2. Buy a lesser known brand. There are a handful of analyzers by Chinese companies that go for cheap. They may be perfectly fine, I just prefer to go with a good established brand if I'm going to invest in one.

3. Watch local auctions. There are a ton of company liquidation auction houses such as Dovebid that sell off large companies test equipment assets. These are great places to pick up gear. The issue with this is that there is no guarantee that the gear works (no one tests it) and if it is a valuable piece it will most likely get bid up pretty high. Packing and shipping can cost hundreds of dollars as well if you are not able to pick up the gear locally.

4. My favorite option. Buy gear whose primary purpose is not a spectrum analyzer, but has an analyzer hiding inside it. A lot of communication analyzers and cell phone test sets have an available spectrum analyzer option. I will look up unusual gear on eBay that seems to be selling for cheap and read the product literature on them. You will be surprised on what you will find. I have purchased both of my spectrum analyzers this way.

My first spectrum analyzer that I bought a few years ago is an HP8922H GSM test set. It is designed to replicate a GSM cell station to test GSM cell phones. It also has a bunch of options included one of which is option 006, a 10Mhz to 1Ghz spectrum Analyzer. You can find these for around $500 or less. Now 1Ghz is great, but you eventually reach the limits of what you can do with it. One of my current projects is building a hydrogen line radio telescope which at 1420Mhz is outside of my analyzers reach. I needed something to at least 2Ghz at this point to test my down converter so I began my search again.

While recently looking at more gear that was available I came across an Anritsu MT8801C radio communication analyzer. Not being familiar with Anritsu as most of my gear is HP/ Agilent and Tektronix, I did a bit of research into this particular model and discovered that not only was it an amazing piece of gear, but much like my HP8922H, it has an option for a 300Khz to 3Ghz spectrum analyzer (Option 07):

Being a communications analyzer it has a bunch of other nice features such as a 300Khz to 3Ghz RF frequency generator, and an RF power meter:


Here is a full span of 300Khz to 3Ghz to my outside wideband antenna:

A couple nice things about this analyzer is that it has a large LCD screen, is capable of displaying a full frequency span, it has a resolution of 1Hz, and a very fast interface. I checked its calibrated accuracy with my HP 8656B RF generator and it was spot on which made me very happy as well:

The additional 3dBm loss above is from the mini-circuits splitter I was using between the generator and analyzer.

Included with this unit was a nice shielded RF test chamber for no extra cost. I can only guess what this had cost new:


It will be a great tool for testing devices within a completely shielded environment from external interference.


So all said and done this complete setup cost less than $600. Quite a steal considering new this was a $30K+ piece of gear. With that being said I'm finally excited to test my down converter further and also get my RF front end for my radio telescope underway.

Thursday, January 19, 2012

21cm Line Downconverter Testing

This project is the heterodyning downconverter I have designed and assembled for radio astronomy purposes. It will be used to convert 1420Mhz Hydrogen Line reception from my 10' dish and feedhorn / LNA assembly down to a more manageable frequency of 100Mhz or so. It consists of a mini-circuits mixer along with a Z-Comm VCO for the LO. This particular VCO has a frequency range of 850Mhz to 1600Mhz, ideally covering the 1420Mhz band. A few mini-circuits low noise mmics are used for amplification along with a mini-circuits low pass filter to filter out the original LO and source frequencies.

Last week after returning from vacation I received my downconverter boards and started assembly. Here is the final assembled downconverter ready for testing:

This downconverter as I began testing has excellent performance. I need to do some further testing and measuring to calculate its noise and gain, but I have been very pleased so far. Because of the flexibility of the VCO, I have had some fun during testing downconverting and also upconverting various frequencies around, the following video show upconversion of FM bands to 900Mhz: