What Has Been Seen
Pulling back the loose half of that sheet reveals a similar fluxy solder job to what we saw in January's TL606 tear-down. The bottom-right trace goes directly to ground and, looking at the solder joints, we can conclude there is one MOV across N-G, one across L-G and two across L-N in this cluster.
The rest of the circuit that cannot be observed from the board edges will need to be figured out the old-fashioned way: continuity testing and intuition.
Traced Out
Traced lines on this image correspond to PCB connections, while dotted lines represent connections through inductors or fuses. Imagine where the components' leads are to figure what they are connected to.
As I suspected, the “isolated banks” are only one inductor apart. More surprisingly, those inductors are wired in series between outlets. I have doubts as to how well the isolation would perform in the presence of a significant noise current.
Introducing Some Noise
How do you test noise filtering or isolation? You inject noise and measure how much of it shows up where it is not supposed to. I'm using a vacuum cleaner as my noise source. Although it's not the most scientific test instrument, it should prove more than adequate for demonstration purposes.
The trace shows the current waveform after adding a 700Hz high-pass filter to attenuate 60Hz current by about 21dB. Our point of interest is the 2A step in 100µs at the red arrow occurring two milliseconds after every zero-crossing due to the electronic power controller.
Rigged For Testing
To measure voltages at each output, probes are hooked up to components inside the isobar. Channel one is connected directly at the fuse, channel two connects to the tinned part of the toroidal inductor's lead, while channels three and four connect to the back of their respective outlets. The blue cap provides a bypass path between scope ground and the isobar's neutral to reduce common-mode noise.
How well-isolated do you think the banks actually are? Let's have a look. The following waveforms are taken at the switching point mentioned on the previous slide, since that is where noise should be at its worst. The traces are offset for enhanced visibility.
First Bank
With the vacuum connected to a first bank outlet, there is nearly no visible noise on any of the four banks. You might even think I got the horizontal and vertical scales or offsets wrong, or forgot to turn the vacuum on. This is exactly what I was expecting. The isobar's 1µF X2 capacitor connected almost directly across bank one eats practically all electrical noise, leaving little to contaminate the other banks strung after it.
Second Bank
Now the vacuum is connected to the second bank, one toroidal inductor downstream from the first bank and X2 cap. As before, the first bank shows negligible noise. But the second bank does have about 5VPP from current change through the inductor. I was expecting it to get transposed directly to the third and fourth banks since they are connected through series toroidal inductors, however, noise actually doubles to 10VPP on bank four. It appears the inductors have a self-resonance frequency of about 100kHz.
How much difference would plugging something in bank four to load it down make?
Second Bank, Take Two
I used a spare computer power supply to provide light loading. As expected, the ringing frequency dropped to 12kHz or so, courtesy of the PSU's 1µF X2 cap in its EMI filter. But the magnitude still doubles between bank two and four.
This may shed some light on why Tripp-Lite mentioned the isobar might not be suitable for all uses. Interactions between reactive components can form resonant circuits, generating unexpected voltages and frequencies that could be problematic for sensitive equipment.
Fourth Bank
This time, the vacuum is connected to the last bank, the one after all three series inductors. Voltage spikes scale proportionally at 5V per bank, starting from negligible noise on bank one and peaking at 15V on bank four, exactly as you would expect from a voltage divider.
My vacuum's current slew rate is about 20mA/µs, which is at least an order of magnitude milder than truly nasty noise sources, so the isobar's inductors could possibly end up generating well over 100V worth of voltage noise from current.
Fourth Bank, Take Two
With both the vacuum and PSU connected to the fourth bank, oscillations similar to the second bank with our PSU on the fourth return, except the waveform gets distorted more severely. Again, the noise amplitude scales linearly with the number of inductors between the reference (bank one) and the other banks.
Isolation?
I really like the isobar's mechanical design, and the use of regular outlets instead of stamped metal strips is also a welcome change.
Electrically, the input filter should do a fine job at filtering line noise and blocking most surge energy. However, the four allegedly isolated banks are a let-down. Performance looks more like you have bank one, and then everything else since the series inductors fail to provide anything I would consider as meaningful isolation. In fact, for the limited testing done here, the isobar would have been better off without them. Tripp-Lite could easily improve this by wiring its toroidal inductors to a common live point instead of cascading them, which looks like a significant oversight to me.
Aside from the design concern above, there is no doubt the isobar has the beefiest filtering and surge protection featured in these tear-downs so far.
Questions? Comments? Suggestions? Leave your message down below or PM me.