16. Diode Bridge Part Five

Here in Part Five you see improvements

Trial 1:

I replaced the amplifier and current mirror transistors with the xxxx transistor. Mounted on a small selfmade epoxy PCB, resulting in higher spikes than before - I am disappointed. The epoxy seems to be a bad choice. Unfortunately these very small transistors can't be mounted on airwires only.

Trial 2:

Fixing xxxx transistors only on wires, a wild connected with different length.

Shorted bridge with transistor xxxx in untrimmed condition at 20Mhz - still too high spikes.

Trial 3:

Using transistor xxxx with small wires soldered with symmetrical length. Also the diode array has now symetrical wires and got more distance to the ground. Transistors were selected on the Curve Tracer, for the current mirror. Results getting more and more better, mounting transistors in such a way seems to bring success.

Improving Shielding:

Bridge Input fixing the first shield.

Bridge Input Shield

Heater Control Shield

Right SMB - Bridge Trigger Input

First trimmer mounted. Copper shield grows up  - it's very important to test every single shield which you fix to see a posible difference. If you change too many steps at once and the result get worser, then you don't know which step caused the fault.

I recommend this:
solder - test - solder - test - solder - test........... this method costs a lot of time, but brings you experience which changes cause which difference.


My first output lowpass filter.
Green resistor 100 ohms with a 82pF styroflex capacitor.
It's somewhere in the 20 Mhz -3dB range, delay is less.

Lowpass Corner Frequency -3dB about 20 Mhz

Lowpass with 10V puls.

10V step settles to 16 bit within 91 Nanoseconds.
This delay caused by the 20 MHz lowpass is tolerable.

Bridge Measurements:

Result after trimming the bridge. Upper trace DAC output with 2V/DIV.

Using a 7A26 in 20MHz position plus 10 MHz RC lowpass filter. The window ON spike is very low now, the OFF spike only one division. Today I am not using my blue P11 phospor 7904 oscillocope, I have also a 7904 with the green standard P31 phosphor. Sometimes I change the scope on the workbench, it's like driving another car. The trace alingnment of this oscilliscope should be readjusted.

Using a 7D20 digital scope Plug-In in the 256 times averaging mode. Same deflection factor settings as above.

The 7D20 has a 70 MHz bandwidth, in comparison with the 20MHz of the 7A26 the spikes are a little higher.

7D20 with 10mV/DIV = 250µV/DIV on the Settle-Node.
The spikes are still ON-Screen, not overdriven.

7D20 with 5mV/DIV = 125µV/DIV on the Settle-Node.
The spikes are OFF-Screen now, overdrive is possible.

Using Schottky diodes:

Using Schottkys.
Changing the diode between the two switching transistors and the settling node diodes.

Center - Settling Node with Schottkys.

Both trimmers mounted and adjusted.

Using 7A26 in 20MHz position plus the 20Mhz RC lowpass.
Both trimmers adjusted. The window-OFF spike is almost gone.

Bridge with a long delay time shows both spikes now. The spike are adjusted in such a way, that the ON spike goes to the screen bottom first, this helps when showing the rising slope of the amplifier. The OFF spike is adjusted to the smallest possible peak. (7A26 with 20MHz + 20MHz RC lowpass).

Even with the more sensitive 250µV/DIV the spikes are still ON-Screen.
(7A26 with 20MHz + 20MHz RC lowpass).

Bridge is ready for 250µV/DIV

Some experiments, I want to replace the external pulsgenerator by a simple Timer IC, saving space on the table, AC line power and the fan noise from the generator.

Changing the Trimmers and more Shielding:

Smaller trimmers and a comfortable shielding. I did some improvements with other transistor positions. The parasitic capacitance of the transistors to the ground leds to different spikes. I reached after many hours a level where I know which transistor cause the direction and heigth of the spikes. Also some influence where the trimmers are soldered on the net, some millimeters more or less and the spike looks different. This experiments took me a lot of time.

Mounting the trimmers close to the ground is terrible and increase the spike too much. That's why the shield around the trimmers is far away from this sensitive trimmers and the bridge nodes.

For example, moving this diode-connected-transistor only a little more closer to the ground, the spike increase some millimeters on the scope. By finding the right position, spikes can be reduced.

A massive ground everywhere is a good basis.

The roof is closed ----- I was very happy ----- I was satisfied with the low spikes and decided to close the roof.

I CHANGED THE RC LOWPASS from 100 Ohms to 220 Ohms with 82pF, a 10 MHz -3db corner frequency. I did it because I want the measure with the 7D20, and that Plug-In measures with 70 Mhz and don't have a bandwidth limiter switch. I also think it's a better idea to reduce the bandwidth direct after the Residue Amplifier and not in the Oscilloscope Plug-In.  The 220 ohms resistor helps also the xxxx amplifiers to isolate the output better from the heavy capacitive 82pF load. The 20pF scope input capacitance adds an additional capacitance.

I completely removed the diode array heater circuit, because there was sometimes a very small HF oscillation, I don't know where it came from. No change to reject, trying many different capacitors to reduce the heater bandwidth, no change. I don't know, may be my unshielded heater wires acting as antennas, after removing the heater it was better. I don't need the heater, the instrument is stable. After burying in cooper, no air movement - the baseline stands stable enough - saves also current, because I thought also about battery operation.

500µV/DIV with 7A26 in 200MHz position plus 10MHz RC lowpass.
The window OFF spike is very low and still very fast.

500µV/DIV with 7A26 in 20MHz position plus 10MHz RC lowpass.
The window OFF spike is almost gone and comes close to the noise level.

250µV/DIV with amplifier 7A26 in 200MHz position plus 10MHz RC lowpass. The OFF spike is still within the CRT. Now you see a very small rising window-ON spike. These small spike can't be adjusted with the trimmers, it was only possible to minimize this small spike with the position of the transistor described above. Also the noise level is low, I guess the shielding helps for less noise.

250µV/DIV on the settle node - 7A26 in the 20MHz position plus 10MHz RC lowpass.
I can't complain about oscilloscope overdrive - excellent.


Impressum und Haftungsausschluss