ferdinand-keil.de

I had some money saved to buy new lab equipment. In the end I settled on an analogue oscilloscope and a power supply. For the scope I bought a used Tek 2225 off eBay. It’s the one that Dave Jones reviewed and recalibrated. I really like this one. The noise floor is lower than the one of my Rigol DS1052E, which is great for measuring audio circuits. Being an old-school scope it has no “Auto Setup” button – so you have to set it up yourself and learn a lot about scopes along the road. One special feature of the 2225 is that it can measure down to 500 µV/div with its integrated amplifier.
When I got the unit it was in good condition. I opened it up and found that a strut inside it was broken. This problem was easily fixed with some epoxy glue, although I could have probably left it as it was. It works like a charm, the CRT still shows crisp bright lines after so many years (the unit was made in 1987).
I already had a power supply, but a special offer from Helmut Singer made me buy an HP 6632B. It has no front terminals, but besides that it’s an awesome power supply. It can measure current down to 2.5 µA and supply up to 20 V, 5 A. Current and voltage can be programmed via a keypad – being a lazy person I love this feature. The sense input is available via screw terminals, which is great at high currents or to offset the burden voltage of a DMM. Oh, and it is programmable via GPIB (or HPIB) and RS232. I still have to build a connector box for this one, but I already tested it with great results.
Narrat1ve also bought one of these and wrote about it in his blog. Mine didn’t need the cleaning but I don’t have the equipment to calibrate it – yet

And now for some pictures.

Some time ago I ordered samples of TI’s TPS61085 step-Up DC-DC converter. I have been looking for a DC-DC converter capable of more than the usual 5-12 V output, and the TPS61085 looked like a perfect match. This great little chip (MSOP and TSSOP packages) can run off as few as 2.3 V and output up to 18.5 V. It integrates an 2.0 A, 0.13 Ohm power switch and offers a selectable 650 kHz and 1.2 MHz switching frequency. The high switching frequency makes it possible to use very small inductors and capacitors, thus reducing the needed PCB area.

After the parts had been lying around for some time I finally had the time to layout a PCB to try them. The PCB layout is based on the layout of the TPS61085EVM evaluation module made by TI and the hints from the IC’s datasheet. The resulting PCB is just 29.21 x 25.40 mm in size, although it could still be made smaller.

I had two of these PCBs made and built one up for an output voltage of 15 V. As I didn’t have the exact resistor values for the feedback divider, the output voltage ended up to be 15.7 V instead of 15 V. I then ran some tests.

Output ripple at low load current.

Rload	550 Ω
Iload	28.3 mA
Vout	15.88 V
Vripple,p-p	17.2 mV

Output ripple at high load current.

Rload	~ 156.7 Ω
Iload	101.6 mA
Vout	15.71 V
Vripple,p-p	33.6 mV

I originally planed to use this IC in a headphone amplifier, so I was curious about the transient response of the circuit. I replicated the measurement pictured in the datasheet to be able to compare the results. Two paralleled resistors were connected as load, one of them through a NPN transistor. A Teensy board was programmed to switch the transistor at 1 kHz and 50 % duty-cycle. The results are shown below.

Transient response.

Rload	360 Ω
	~ 78.3 Ω
Iload	44 mA
	~ 200 mA
Vin	4.99 V
Vout	15.94 V
Vtransient,p-p	~ 200 mV

Of course I also determined the efficiency of the circuit. At a load current of 44 mA the efficiency turned out to be 82.75 %. Not too bad for a first revision. The complete data is shown below.

Rload	360 Ω
Iload	44 mA
Vin	4.99 V
Vout	15.93 V
Efficiency	82.75 %

The TPS61085 is a very capable little boost converter IC with many possible applications. As I want to use it in an audio circuit, low ripple at high frequencies is key. It also uses very little PCB area and matching inductors are tiny and very cheap. From a 5 V USB power supply this chip, together with an inverter can supply +/- 15 V for precision OpAmps.
With an output ripple between 17 mV and 33 mV the circuit as is would probably need additional regulation if used to power a tube or Class A circuit. In this case the high ripple frequency is a problem, as most common regulator ICs have pretty low attenuation at high frequencies. A simple LC filter might suffice to attenuate the ripple, but I haven’t tested that yet.
As I want a power supply with a positive and negative rail, I am right now looking at the TPS65130, which integrates two converters in one IC. Building an evaluation PCB for this chip is next on the list.

These days I swiped the Windows installation from a Shuttle XPC system and set it up with a clean install as I plan to give it away. While I was cleaning the mainboard from dust, I found the single SRAM module to be sitting in the slot DIMM1. On most boards a single stick of RAM is expected to be put in slot #0, so I swapped the module over to the first connector. I then updated the BIOS and installed Windows XP SP2, but not without difficulties. The installation would not finish with an AGP graphics card connected, and even after I removed it the network adapter was neither recognized by Windows nor a Linux system.
Downgrading the BIOS and clearing the EEPROM did not help, so in an desperate attempt to restore the LAN adapter I plugged the RAM module once again in slot #1. And you know what, Windows boots up and recognizes the adapter just fine. This is not mentioned in the mainboard manual and installation guide so I had to find out the hard way.
So if your Shuttle FS51 v2.0 (aka SS51G) acts weird try to change the RAM slots.

---

Update 25.08.2011
Well, as it turned out the RAM wasn’t the problem. After I reinstalled Windows XP on the machine the network card did not show up in the device manager. In a desperate attempt to restore the network connection I measured the BIOS backup battery and voilà, it was empty (down to 1.5V). I replaced it with a fresh one and the network card magically appeared in the device manager. Everything seems to be back to normal now so I can finally finish the installation process.

---

Update 27.08.2011
The network adapter disappeared again! A CMOS clear brought it back for some time, but then it disappeared again. The CMOS clear should be done after every BIOS update by moving the on-board jumper to its second position and back after several seconds – lesson learned. After checking the power supply (all good), the caps on the mainboard (also good) and a lot of hair tearing I finally came back to the BIOS settings. I hadn’t changed the “Reset Configuration Data” switch before so I gave it a try. And the network adapter finally came back. It even stayed after I changed back the switch to Disabled. I don’t now how long it will take until my RTL8139 goes back to its happy place but for now I’m confident that I solved the problem.
I read about the meaning of this BIOS setting and what the Extended System Configuration Data (ESCD) is, and it seems plausible that this was the problem. Time will tell…

---

Update 28.08.2011
I give up. The network card has been and gone. Under the screen of night it left and I have been unable to bring it back. From the looks of it something is seriously messed up with this PC, but I won’t bother even further.

R.I.P.
RTL8139
2003 – 2011