Latest Entries »

Tenergy Li-ion 4400mAh Roomba 400 Battery Review


Several people over the years have found my blog posting on why not to use Lithium battery packs in Roombas.  Some people have taken the information at face value, others have argued with me in the comments and tried to claim my views on the subject are not valid.

Some of the folks over at Tenergy also noticed my article and decided to send me a battery to prove to me they have resolved the issues I was citing.  Here is a link to the battery in question:

In the interest of full disclosure, in consideration for the battery, they only asked for my honest review along with a link back to the product to allow people to find it.


At this point in time, I have been using the battery lightly for about a month.  I use my roombas somewhat sporadically.  This may actually shorten their life due to lack of cycles and a poor factory charger circuit.  So far, this pack seems to be holding up just fine.  One thing I should point out that is different about this pack from the other lithium packs I used in the past…  When the other packs ran low, the Roomba light would start to glow orange and then suddenly the roomba would die in place.  In other words, it never had enough charge left to bring itself back to the base.  The batteries low volt protection cutout would kick in and save the cells from damage.

This pack appears to work like a typical NiMH pack.  The light eventually goes red like it should and the robot stops moving but the power remains on.

As far as runtime, I’m using the pack on hard surfaces.  I haven’t paid close attention but it seems to go for a couple of hours.  One day I will time it and update the review, for now though, I wanted to post my initial thoughts.

Similarly, I will obviously need to reserve my judgement on longevity until the pack starts to noticeably deteriorate.  That could be months or could take over a year.

So far, if I had to pick one complaint about the pack, I would say that one of the retainer clips is unable to snap into place.  This is a pretty minor issue that I could probably fix with some sandpaper or a knife but I shouldn’t have to.  I’ve cleaned the area under the battery and pushed on it pretty hard but it just won’t snap in there for some reason.


For now, I will say that the price is currently $68.99.  This is marked down from the typical $90.  The Tenergy packs for the 400 are $25 on Amazon.  So what do you get for 3x the price?  I’m not sure yet.  Only time will really tell there.  The pack is certainly lighter than a NiMH pack so it’s putting a little less wear on the robot.  It seems to last similarly to some of the best NiMH packs I’ve had.  It’s not exactly an all-day battery pack however.  In the coming months, I’ll try to monitor the health of this pack a bit more closely and possibly more scientifically.

My first lithium pack for the Roomba was $150.  When it prematurely cooked, I was not pleased.  At $69, if I got a year out of this one, I probably wouldn’t complain too much but then again, if it only lasts a year, why pay any premium on it at all?  Let’s hope that I’m able to favorably speak of this pack for years to come as “the best Roomba battery I’ve ever had”.

Dead Hypex SMPS600 V3 Power Supplies


A couple of years ago I decided to build myself some class-d amps based on the Hypex NC400 modules.  I had also designed some custom billet aluminum cases but I knew they were going to take a while so I put them in temporary enclosures.  I had noticed during the summer months when the weather was warmer, one channel would click off.  I finally figured out it was thermal shutdown and so I turned the case over and put some metal objects on top to act as heat sinks.

Fast forward to a couple weeks ago, I needed to turn my equipment off to make some changes to the setup.  It took me overnight to make all of the changes and when I went to plug the amps in the next day, neither would turn on.  No signs of life at all.  No clicks, no lights, nothing.  I was a little confused that both could die at the same time.  They are in separate cases on opposite sides of the room but when I popped them open, I reached for my cap wizard and started testing out some of the caps in the power supply.  Turned out several were cooked.  Some did not even move the needle on the cap wizard.  To my suprise, even one of the largest caps, a 820uf @ 200v, had drifted far out of spec.

I decided I would just replace all of the caps on the power supply boards because I wanted to make sure they were perfectly matched and I figured they all probably have a shortened life span from the heat they incurred.  My first task was to remove and map out all of the caps.  To get to the 2 largest caps on the board, the large heatsink HS1 needed to be removed.  In order to remove that large heat sink, R2 needed to first be removed to access the screw holding D43 to the heat sink.  After I got that out of the way, then I could  remove and catalog the caps.  This board is not particularly easy to work with.  The holes are drilled with very small tolerances compared to most.  This means you really need to get all of the solder out of the holes or the new components won’t go in easily.

4x 820uf 200v (C1, C15)
16x 220uf 35v (C3, C4, C5, C6, C7, C8, C13, C14)
4x 22uf 63v (C22,C26)
2x 470uf 50v (C9)
12x 100uf 100v (C10, C11, C12, C19, C20, C21)

After that, I ordered the caps for about $40 shipped and replaced them when they showed up.  So far so good.  Everything seems to be working again.  Lesson learned.  Even though these amps use a very small amount of power and run very cool for their power outputs, cooling is still super important.  I would suggest having plenty of cross-ventilation in your cases.  Make sure you leave adequate distance between the amp and power supply.  Also make sure your NC400 module is attached to a larger and thicker piece of metal to properly cool it.  Apparently a smallish piece of light gauge aluminum is not adequate when in an enclosed space.  If you need to replace your caps, ONLY use 105 degree varieties and use the ones rated for the most amount of hours at that temperature.

NC400 & SMPS600

Sharp Image SI-727R-DS Repair Log

This monitor has been on my desk for months because I’ve been avoiding it but I’ve decided that it’s taking up too much space so I’ve decided to start digging into it again.  I’m going to document my repairs here in hopes that it will help someone else out to see my troubleshooting process.

The symptom of the issue is no picture.

The problems are several.  The monitor is switchable between standard and medium resolution hence the “DS” in the model name which stands for dual sync.  When I attempt to run it at 15k, things seems somewhat stable.  There is no picture but when I test B+, it will sit there right around 130v which seems pretty reasonable.  Testing with the high voltage probe I get about 130v on the anode which tells me something in the flyback/HOT area is not working correctly but I don’t have a schematic so I’m flying blind on it.  When I hit the base on the oscilloscope, I get a reasonably nice looking square wave that seems to be at the correct frequency and looks happy.

Oscilliscope reading (base)

So that makes it appear that the driving waveform is ok.  When I hit the collector, I get this:

Oscilliscope reading (collector)

That does not look good.  Theoretically I should be getting voltage spikes that go well of the screen when the HOT grounds the flyback.  The scope is set for 50 volts per div and I’m barely covering two of them so something is clearly wrong here.

When I switch over to medium resolution (25k jumper) things get much worse.  The B+ pops on and starts at about 140v, then it keeps climbing and the monitor starts getting noisy to the point where I’m concerned and then I shut it down.

I’ve tested the flyback with a ring tester and it looks good but perhaps there is a different problem with the flyback that cannot be diagnosed with that tool.  I know for sure the HOT is good, I have several of them and they all test the same.  When I pull the HOT out of circuit, both the 15k and 25k mode act the same and give the same B+ voltage of roughly 130v.  The only difference is the waveform reading on the base of the HOT.

Mattel Aquarius Composite Video Modification

During the holiday season here I had some time to tackle some of the projects that have been on my todo list for ages.  One of these projects was improving the video output from my Aquarius.  Last time I hooked it up, the picture was pretty horrible.  It may have been a flaky RF switch box but an RF modulated video signal isn’t really ideal in the first place.  Searching around, I found some schematics for the Aquarius so I thought this project would be a piece of cake.

Here is the offending RF modulator

Here is the offending RF modulator

It appeared to be a simple matter of removing the RF modulator from the board and then grabbing pin 1 and feeding it to a RCA jack and feeding pin 3 to another for sound.  When I did that though, I got video like this:


Wrong colors, smearing, illegible text.  Yuck!

Wrong colors, smearing, sync loss, illegible text. Yuck!

It was at this point that I realized the project might not be as easy as I first thought.  I decided that the problems was a weak signal so I set out to try to amplify it.

I was a little puzzled as to why such a weak signal worked fine for the RF modulator but not for the TV.  When I looked at the signal on a scope, it looked fine:

This is the unloaded signal straight out of the TEA1002 chip with the two resistors in place.

This is the unloaded signal straight out of the TEA1002 chip with the two resistors in place.

But then when I put a resistor across the signal to load it down, it squashed down to nearly nothing.  Apparently the RF modulator doesn’t have much of a load at all since it has no 75 ohm cable termination to deal with.

I unsuccessfully tried a couple of single transistor emitter-follower circuits such as this NES Video Booster circuit that I found.  While it improved the signal, it didn’t entirely fix the problem.  My friend suggested a purpose built video buffer chip such as the MAX4090.  Oddly, he had a roll of 200 of them laying around that he never found a use for.  I hooked it up on my breadboard to check it out:

Breadboarding an SOT23-6 packaged chip was a small challenge but certainly not impossible.

Breadboarding an SOT23-6 packaged chip was a small challenge but certainly not impossible.

That's more like it!  (Ignore the moire pattern from my cell phone cam)

That’s more like it! (Ignore the moire pattern from my cell phone cam)

Success!  That worked pretty well.  I started with the reference circuit found in the MAX4090 datasheet but found that it worked best with only the one cap installed on the output.  I omitted all of the resistors from the circuit and the decoupling cap on the vcc.  Just in case though, I designed my PCB off of the reference schematic:

MAX4090 Breakout Board Schematic

MAX4090 Breakout Board Schematic

Here's my Kicad layout for the breakout board.

Here’s my Kicad layout for the breakout board.

Here's the breakout board installed in the Aquarius.  Fits where the RF modulator was.

Here’s the breakout board installed in the Aquarius. Fits where the RF modulator was.


This is how it should have looked from the factory the day it was shipped!

In case you are wondering about the red +5v line, I got that from the bottom of the board.  The 5v regulator line is very clearly marked on the solder side over near where the three wires from the regulator go under the metal shield.  I just followed it as close to the former location of the RF modulator and grabbed it there.

S.T.U.N. Runner Random Resets


The opportunity finally presented itself for me to buy a full-sized arcade game.  I suppose I could have picked a smaller one but I kind of bought it on a whim.  The game?  S.T.U.N. Runner.  It’s a sit down “cockpit” style racer with no gas pedal.  It’s a game that I used to play all the time whenever I would see it back in the early nineties.  This particular copy was obtained from the B&I Amusements auction when they shut down in 2013.  I was pretty disappointed to see them go but I thought this was a reasonable souvenir.

The game was not perfect when I got it.  The cabinet and plastics were not bad at all but the monitor was so dim it was totally unplayable.  A recap and rejuv made the monitor really nice again.  There was some other random small issues such as a cracked handle on the steering yoke.  Luckily there is a stockpile of over 5,000 of these at Suzo-Happ.  The Dallas battery powered RAMs which kept high scores were also shot.  Luckily a new comparable part is still manufactured.  I ironed out a most of the small issues but then another one popped up.  The game started resetting randomly.

At first I sort of ignored it because I figured it was just a ghost in the wires or something but then it started making the machine unplayable.  I finally got so frustrated that I ordered a new board set off eBay.  I figured that whatever was making this happen HAD to be on the board.  I put the new one in for a bit and it seemed to fix it for a bit but soon it started again.

Surely the only other culprit could be the power supply, right?  Wrong.  I tried a different one and the problems still continued.  All the was left to look at were the connectors and wires.  I started furiously blasting all the molex connectors with DeoxIT.  After waiting the recommended two minutes and powering it back up, I was hoping to see the end of the resets but did not.  As a last ditch effort, I decided to start unplugging harnesses because I figured something external had to be killing it.

Turns out that not much experimentation was involved after all.  I unplugged the harness going to the volume control and service switch.  I put my son on the game to playtest it for me and he decided he wanted sound again which doesn’t work without the volume obviously.  I plugged it back in live and what do you know?  It reset, right there.  Eureka!  I pulled the service switch and volume bracket and stuck my VOM probes on the switch.  It read overload at first which was fine but then all the sudden started bouncing between 100-300 ohms.  No wonder this poor game was confused.  This switch started conducting if you sneezed on it.

The date ended with me blasting the heck out of the switch with DeoxIT and me needing to list a perfect second board set on eBay.  :-)

Neotec NT-27E catastrophic failure mode

It’s hard to tell what went first…  Was it the flyback?  Maybe it’s the mylar with the guts oozing out the side.  The weak green gun on the crappy Zenith A68AGD01X tube?  Perhaps it was a perfect storm but my sneaky suspicion is still that something went first before the rest.  The frusteration came however when I replaced the flyback, all the burnt components and all of the caps and still got no signs of life.

bad electrolytic


bad mylar

scorched resistor


That signaled the time to bring in the rejuv.  I haven’t used a rejuv in years so this was pretty exciting.  One of my friends just picked up a Sencore CR70 off of eBay.  It’s a really sweet box.  Way nicer than the one I used at All Repair back in the day.  This one only has 5 sockets.  You set the wiring diagram but using a series of switches on the unit itself.

Anyhow, we fire up the rejuv and found the green have pretty lousy emissions.  Worst yet, the cutoff didn’t even register on the scale until we cranked the negative bias from the recommended 68v down to somewhere in the 35v range.  We hit it with the auto rejuv 3 times and then with the manual cycle.  This brought the emissions back to the level of the other guns but not the cutoff.  There was a little bit of improvement but it’s a clear sign that this tube will need to be replaced sooner than later.

After rejuving however it still didn’t work.  We had heater and high voltage though so we were a bit perplexed.  We ended up checking the G2 voltage live in circuit.  The most we could squeeze out of it was 190v or so.  This was way too low given that a similar chassis with the identical tube a few games downw was putting out 290v to get a reasonable picture.   We had the Neotec NT-27E service manual so we gave the schematic a once over.   It APPEARED that our fault may have been in the brand new flyback but my friend didn’t want to give up so easily.  he decided to check every pin of the CRT to ground.

scorched socket

Low and behold, G3(focus) was reading less than 1M of resistance to ground.  Not particularly good.  A closer look at the socket revealed that it had gotten really hot.  We picked at it until it opened up and ended up with a handful of carbon residue.  It turns out there is a spark gap inside of this socket.  It had started arcing which produced some carbon.  This led to more arcing until the focus had shorted enough to ground where it was also dragging down the G2 voltage substantially.

For the moment, we yanked the spark gap and soldered the focus wire directly to the focus pin.  Needless to say it fired right up.

Not sure how long this would have taken me to discover myself but it was a darned good find on my friend’s part.  Very impressive.

Fixing the AT&T 6300/Olivetti M24

I was at the computer recycler a couple of years ago and this machine caught my eye.  It was a very different looking machine compared to others of it’s era.  The case has a sidecar for the hard drive unit which makes it look even more odd.  When I bought it, the monitor and keyboard cables had been cut in two.  Instead of reaching around with a screwdriver to properly remove the cords, someone just decided to slice the cables.

I fixed that at least a year ago.  I popped open the monitor and meticulously sorted out all of the wires needed.  You might think that would be fairly trivial but this monitor uses a DB25 connector.  I didn’t know it was actually a color CGA (sort of) monitor until I popped it open to do the sorting.  After a couple hours of sorting, I finally got it all wired up.  I plugged it in and saw this:

garbled olivetti screen

At that point it became obvious why the computer was discarded.  I got busy and stuck the machine aside.  Recently, in a fit of cleaning, I stumbled on the machine again (literally).  I figured it was time to do something with this box or toss it.

I fired it up to the same old familiar screen.  Next, I pulled ALL of the cards.  I wanted to make sure those weren’t somehow acting poorly.  This machine is VERY unique.  The video card is also the backplane.  The motherboard lives on the bottom of the case and is connected to the expansion slots via the video card.  When I get the machine fully running, I’ll post more pictures that will better explain this.


Once I had room to maneuver, I hooked up my logic probe and logic pulser and actually managed to locate a schematic for this beast.  I started out just probing around randomly until I find some interesting signals.  Occasionally I would hit a chip with the pulser.  Usually the effect was just some noise in the video signal but then I started getting closer to the right section.  I found a chip to probe where on of the pins would pulse every time the flashing line on the screen changed.  AHA!  Must be close now.

I looked some more at the schematic and found the character ROM where the default font is stored.  I probed around on it and decided that it must be bad since it’s the only thing I can’t replace… so I thought.  I took a look through my junk box and found the original CGA video card that came with my PCXT.  Low and behold it had the exact same EPROM on it.  I figured that was random luck though and there would be no way it would work even if it was the same chip.

I decided to swap it and by some form of magic, it did in fact work.  The problem still remained exactly the same though.  I decided this was good though and moved on to other areas.  I decided to eavesdrop on the four TMS4416-15NL DRAM chips.  I figured that the signals in and out of them should all be similar so I compared the signals.  I found that one of the chips had a couple of dead data out lines.  Since all of the address lines appeared to be doing what I would expect, I determined that this one chip must be the problem.  I also decided to hit a couple of the lines with my pulser.


This reaction proved that I was on the right track.  Once again, I went searching on the IBM CGA adapter and found an exact match for the ram chip I needed.  Thankfully, the ram on the IBM board is socketed.  Unfortunately this is not the case with the Olivetti board.  I had one socket that fit in my parts box though so I popped that onto the Olivetti board and popped the chip into it.  After reassembling enough of the machine to test the board, I was eventually rewarded with this:

Olivetti Working 1.43 BIOS screen

This is probably not the end of the problems with this machine.  Far from it in fact but it’s an interesting machine in good condition so it’s well worth the effort and the digital trouble shooting skills I’ve gained from this experience are well worth the $10-$20 I paid for the machine in the first place.

G07 mistakes

The Electrohome G07 is a simple monitor…  At least that’s what I keep hearing and seeing.

As simple as it is however, I have not had a lot of experience with the G07.  The majority of my experience has been with 1990’s computer monitors and more recently 1990’s arcade monitors such as Wells Gardner and similar.  Compared to the later monitors, the G07 is completely foreign.  For instance, the horizontal output transistor is mounted off of the board on a heatsink.  It’s insulated from the heatsink to keep it from grounding out because the outer case of the transistor is one of the conductors BUT, this conductor does need to go to something so Electrohome uses a bracket with a wire on it that the mounting screws go in to.

The problem I had ran into was that NTE had included additional insulators in their replacement transistor kit.  I had incorrectly assumed that more insulating is better.  The problem being that it broke electrical contact from the heatsink AND the back bracket which resulted in an open circuit, no high voltage and a dead monitor.

This is a cautionary tale.  The last 3 monitors I have not been able to fix and needed additional help on were due to improperly mounted HOT’s.  Some day I’ll learn this lesson.

Another note about the G07, the hot may appear to test as shorted because of an internal diode.  Always make sure to test from the pins to the case.  Testing between the pins looks like a short.

Wells Gardner WGM2775 Missing Colors


For months now Vapor Trx has been driving me crazy.  Every time I looked at it I wanted to sort out the issue.  The problem is that sometimes all of the colors were there and it would work great.  Then usually red would flake out and sometimes green.

I had pulled this board out and reworked it over a good bit.  It had a lot of lifted traces from heat issues and other solder issues along with needing a few caps.  I had done all of this, shoved it back in and it worked great for about a day.   I finally got the chance to dig into it a bit more.

I started by re-reworking the neck board since this is the most likely place for a failure with a color channel to occur.  I popped it back in and it was once again fine for a bit and then it died.  I popped it out and felt that there are 3 color driver transistors that looked very suspicious.  Those were the main spots where the traces had lifted due to heat and they all have crappy clip-on heatsinks that were kind of loose.  I decide to see if the part was bad or if I was having connection issues on the PCB.  I decided to swap the red drive transistor with the green one.

When I tossed it back in the machine, it worked perfectly until I came back the next day.  Luckily it was once again missing a color, but this time green.  So now I could be certain that the transistor itself was failing under load.  I stopped by Vecto, grabbed 3 NTE198’s and popped them in.  I tightened up all the heatsinks and fired it off.

I don’t want to jinx it but I’m now pretty certain that this one is fixed.  Hopefully people start playing it now that the screen actually looks good…

Sharp Image SI-727R-DS fixed!


Well it’s my fault.  I should have known better and checked this but I had tossed in the incorrect horizontal output transistor.  I had purchased a box of 5 replacements off of eBay for a Wells Gardener 25K7401.  That monitor however had a silicon insulator pad that kept the back of the HOT insulated from the heatsink.  The SI-727 however, does not.  When I stupidly installed that part in the unit, it was shorting out and giving me tick-tick-tick noises because the switching power supply was trying to fire up but went into protective shutdown.

Once I fired the monitor up, the picture looked terrible.  Looking into the cabinet though you could tell an amateur did the harness hack.  He twisted wires together and poorly electrical taped them.  This was not the problem but I fixed it.  Ultimately the problem was that the signal ground came detached in the process of fixing the monitor.  For the life of me, I could not find the proper place to attach this wire but ultimately I found one suitable at least.

The monitor still isn’t quite right.  The picture grows and shrinks slightly when going from light to dark scenes(monitor bloom) and the image does not quite fit the screen.  As annoying as this is though, the game is at least playable now.Lessons learned?

Always make sure none of the legs of the HOT have continuity with the heatsink unless of course it’s designed to do so and if you picture every looks inexplicably bad, check your grounds.

Powered by WordPress. Theme: Motion by 85ideas.