Revised July 11, 2006
Voltage dividers provide an unambiguous means of measuring the high voltage applied to an x-ray tube. Most voltage dividers have an accuracy better than 1%, if calibration is in date. They can be used improperly, which will negate the inherent accuracy of the instrument. In its most basic form, a high meg... ohm, high voltage resistor is connected in series with a resistor of much lower value. The divider is then connected to the inputs of the oscilloscope, or to the 10 x scope probe, depending on the manufacturers instructions. For example, the GE high voltage "bleeder" has a 100 meg...ohm resistor array connected in series with approximately 101,000 ohm viewing resistor. When a 10 x, 10 meg.. ohm oscilloscope probe is connected to the output terminals, the oscilloscope will present the waveforms at a scale factor of 1 volt 10 KV. The scale factor at the binding posts of the bleeder tank is 10 volt=10 KV. The Machlett HV-1 divider can be connected in several ways, direct or through a probe, leading to large errors if not set up properly.
I have seen some confusion out there, and it is VERY IMPORTANT to be clear about these points:
#1: The H917 / GiCi 2000 divider, which is also sold by Fluke Biomedical has two modes of operation from the switch on top. One mode is for direct connection to an oscilloscope, without probes... just a BNC cable...any impedance. When the switch is set for 10,000:1, it is set for a 1 meg ohm load. The same applies to the HV-1 divider from Machlett-Varian-Radcal. You MUST set two switches one to 10,000:1, and the other to 1 MEG OHM. 1 Volt out of divider is equivalent to 10 KV.
#2, If you are connecting the H917/ GiCi 2000 divider to a digital volt meter for either peak capture or steady state reading from a long exposure, use the 1000:1 range which is set for 10 meg ohm load. Fluke 87 and other assorted pocket DVM's are usually 10 megohm input resistance on the DC volt ranges. Lab meters such as the Fluke 8840, 8842, etc are 10 meg with the 200 volt DC range and higher. When using the HV1 divider be sure the switch is 1,000:1 and 10 meg ohm. Do not use any other combination of the switches other than those described in paragraphs #1 and #2, because they are not separately calibrated and could have errors in excess of 1%.
#3. The older Nuclear Associate bleeder 07-476, which have NO switch on top h ave a ratio of 100,000:1 and a very low output resistance. The low output resistance is not affected by any scope or meter. These dividers do have significant high frequency overshoot.
It is very important to connect a voltage divider to an oscilloscope properly. Failure to do so can cause large errors. The GE divider and the Machlett HV-1 each have their own "traps".
Further discussion...my original text follows
"...the following items are required: 1) Cl5l5A Bleeder, 2) Two 5 ft. HV cables with standard termination at each end. 3) One dual trace oscilloscope capable of algebraic addition. Tektronix model 46 or 465 Model UC are preferred.... 4) One 30 ft. twisted pair cable, low capacitance 2-conductor, shielded. Belden No. 8422 (18 mmfd/ft) or equivalent. A 17 ft. length of cable must be used to obtain proper calibration on AMX-3. This permits locating the scope away from the area of radiation for safe viewing. Never use the bleeder without this cable as it is part of the calibration!"
(Ref 1) General Electric Co., Direction 13288G, "Connection... Application"
GiCi experiments indicate that the best frequency response of the GE divider can also be obtained by connecting one 470 pf. ceramic capacitor, 500V, from anode binding post to ground, and a second like capacitor from cathode binding post to ground. This provides the equivalent capacitance of the 30 feet of cable. The scope probes can then be connected directly to the terminals, as many technicians are now doing....but without the capacitor. Failure to have the correct equivalent capacitance will result in the divider showing large overshoot in pulses or in high frequency ripple. We have been making this recommendation in our application notes for several years. ( Use the capacitor or the twisted pair, but not both.)
Nuclear Associates Voltage Dividers are relatively easy to use and trouble free. They have a 100,000:1 ratio and connect directly to the oscilloscope input terminals. (1 volt= 100 KV.) . This divider is frequency compensated, and has a 100 meg ohm internal resistance. Because of its low output viewing resistance ( 1 k ohm), variations in load resistance and cable capacitance of little effect on its performance.
Tech Talk:
It is important to understand the construction of the voltage divider you are using in order to properly use it. Voltage dividers differ in internal resistance which vary from a low as 60 meg ohms to as much as 1000 meg ohms ( 1 gigohm). See table 1. Some voltage dividers are frequency compensated to a greater extent than others.
When using a voltage divider, it is important to match the resistance of the readout device such a oscilloscope direct input or probe input to the divider. The GE high voltage bleeder type C1515A is designed to be connected to a 10 megohm load. If it is connected by mistake to a 1 meg ohm load, then the user will see the output drop by 9%, because the output load resistor will shift from 100k ohm, to 100k ohm in parallel with 1 meg ohm, or 90.9 K ohm. The mismatch error with the HV-1 divider is significantly larger, because of the higher resistance of that unit. The HV-1 divider has several modes of operation which can lead to error and confusion. Because of its high resistance, small mismatches in viewing resistance can cause large errors.
The Nuclear Associates divider tank has a 100,000 :1 ratio. It has a 1000 ohm viewing resistor, so whether you are using a 1 meg scope, or for some reason a 10 meg scope probe, the output will change no more than .1% .
What makes a divider frequency compensated? In the real world of voltage dividers, the divider resistors have some stray capacitance across them, as well as stray capacitance to the walls of the container. To frequency compensate the divider, it is important to reduce the stray capacitance, as well a compensate the divider network. Elementary circuit theory for a voltage divider requires that the following equation be met, where R1 is the high voltage resistor(s), C1 its compensating capacitor(s), and R2, C2 the viewing resistor and capacitor.
Simple 10 x scope probes have a trimmer capacitor, C2, to adjust the probe to a scope, and so do high voltage bleeders. The Dynalyzer is also highly frequency compensated, but uses additional compensation as well.
TABLE 1
|
Model |
Divider Resistance |
Ratio |
Load Resistance |
Freq. Response 3db* |
|
General Electric C1515 A |
100 meg ohm |
1000:1* |
10 meg ohm |
2.5 kHz+ |
|
Nuclear Assoc. 07-476DT |
100 meg ohm |
100,000:1 |
1 meg to 10 meg |
15 kHz |
|
Machlett HV-1 |
700 meg ohm |
10,000:1 |
1 meg to 10 meg |
less than 100 kHz |
*10,000:1 when used with 10X scope probe
** based on GiCi measurements. Note: This does not mean that the frequency response is flat for any of these dividers. 3db point is where the output of the sine wave of that frequency is down by 50%.
+There are no internal compensating capacitors, however the load capacitor and wiring is critical.