Power supplies
Hewlett Packard 6209B
For testing anything between piezo drivers and tube gear, you need more than your standard workbench supply. Bench supplies in the 300 V range are not common, and an old 6209B could be the answer. It delivers up to 320 V, though only at up to 100 mA, but this is sufficient for many purposes. The serial number prefix spans over almost two decades, so front panel details vary, and so do the prices. For higher currents, or a newer model, consider the Instek GPR-30H10D, or perhaps the HP 6035A. |
SAFETY WARNING!! -- Better check out your 6209B
Does this concern a design flaw or a production error, not something you would expect from HP? Or, perhaps more likely, is this an example of a badly made repair? The issue is that the screws for the power transistor are so long, that they actually are in contact with the bottom plate. The marks on the picture on the right tell the story. The result: The positive rail is connected to the chassis on this unit, and touching the negative output and the chassis at the same time could be lethal. Note also how the pins of the power transistor have not been shortened sufficially. The unit here has serial prefix 2409A, one of the late models, so I would never expect this to come out from an HP plant. By looking at pictures on the Internet of other 6209B units, it appears that the screws on this particular unit are oriented upside down, and the pins of the transistor are definitely longer. I suspect the power transistor was indeed replaced at one point. Good thing I always inspect and check gear before applying power... After I took the pictures I reversed the orientation of the screws, I shortened the pins of the power transistor, and I then added adhesive rubber feet to prevent the chassis from bending down and touching the screws. May I recommend that you check your 6209B and do something similar, if required?! |
Reducing the peak-to-peak noise voltage
The 6209B is a linear power supply with an SCR pre-regulator. This combination provides a fairly low dissipation, but the risk is a feed-through of the transients from the SCR turning on. The RMS ripple and noise of my 6209B is only about 0.3 mV in a 20 MHz bandwidth, fairly independent on the load, and with a good margin to the specified 1 mV. However, the peak-to-peak noise is much higher, about 30 mV. Though still below the specified 40 mV, it makes you wonder if anything could be done about it. The digital oscilloscope screen dump reveals the spike originating from the SCR pre-regulator. The position relative to the mains voltage varies with the voltage setting, and to some degree on the load. The picture below shows the output after a modification of the 6209B plus an external filter, at 160 V output voltage, with 20 MHz bandwidth setting for the oscilloscope TDA340A. The peak-to-peak voltage was reduced from 17.4 mVpp to 3.14 mVpp. The peak-to-peak voltage for the unmodified 6209B increases to 30 mVpp when using a 100 MHz bandwidth setting. The measurements were carried out through the following network below to avoid ground loops. The transformer used, the Murata DA101C, has a measured frequency span (-3 dB down) from 2.2 kHz to about 100 MHz, in an 50 Ohm environment. Due to the source and sink impedances, the peak-to-peak values shown on the left should be multiplied by two. Note: In case your would like to use the network, turn the 6209B on and off with the network connected to allow a controlled charge and discharge of the 1 µF capacitor. |