RAT Noise Tester DISCLAIMER: All this material and accompanying files are (c) Copyright, 1997-1998 Steve Bench. All rights reserved. Permission is extended to build and use this equipment for your own use. You may not build-for-sale-or-other- personal-gain without royalty arrangement with the author. You may freely post this information on your web site; give the author credits, though. WARNING: This equipment, like most vacuum tube (valve) equipment, uses potentially lethal high voltages in its operation. The author is NOT RESPONSIBLE for any accidents using this equipment. Build AT YOUR OWN RISK. PURPOSE: To be able to select vacuum tubes (valves) for noise performance under many different operating conditions. This is a stand alone device, and includes it's own metering (a 0-1 mA meter). You can use the raw power from the RAT Tube Tester, or build the one I did for this project. This file assumes stand alone operation and details the power supply AND the Noise Instrumentation. SCHEMATICS: The Power Supply is listed as SUPPLY2.GIF. The Noise Instrumentation itself is shown in NOISEP1.GIF. There are three photos that accompany this project. They show the basic noise tester (NOISE1.GIF), the "top deck" showing the switches and tube sockets (NOISE2.GIF) and an "inside shot" showing the shielding around the high gain preamp that gives the tester its sensitivity (NOISE3.GIF). SPECIFICATIONS: Tests devices socketted as 9A/9AJ (12A*7, 12BH7, 6CG7, 6DJ8, 6922 etc) 9H (5687 etc) 9AD/9CQ (5879, 6267, EF86) 9V (417A/5842) 8BD (6SL7, 6SN7 etc) Plate supply voltage: Variable from 80 to about 180 volts, up to 45 mA. Plate load resistor: 8 switches: 1k to about 256k in 1K steps. Cathode bias resistors: 8 switches: 10 to 2560 ohms in 10 ohm steps. Types of noise monitored: Flat (20Hz to 20kHz) or RIAA equalized. Heater voltage is variable from about 4 to 6.2VDC. Monitoring: An 0-1 mA meter is configured as: 1. Heater voltage (0-10V). Combined with heater resistance POT, allows heater current to be calculated. 2. Cathode voltage (0-10V). To allow you to see the operating point. 3. Anode voltage (0-200V). Ditto. 4. Noise. Logarithmic scale from about 2 uV to 300 uV full scale. This is noise at the plate. Equivalent input noise must consider stage gain. Residual noise is 2 to 10 uV, depending on the load. Additional features: Includes a 60 Hz notch to aid in eliminating hum from the noise reading. Measures low noise pentodes in triode mode. Controls: On/Off, plate voltage, heater voltage, meter function. (On front panel) 8 plate load toggle switches. 8 cathode bias toggle switches. Side A/B (for dual triodes). Type of noise selector (flat/RIAA). 9A/9AJ configuration toggle switch. 9AD/9CQ configuration toggle switch. Meter Calibration. Meter says. Heater Volts Cathode Volts Plate Volts Noise 0 0 0 0 < 2 uV .1 1 1 20 < 2 uV .2 2 2 40 2 uV .3 3 3 60 4.6 uV .4 4 4 80 8.5 uV .5 5 5 100 17 uV .6 6 6 120 32 uV .7 7 7 140 60 uV .8 8 8 160 105 uV .9 9 9 180 180 uV 1.0 10 10 200 300 uV CIRCUIT DESCRIPTIONS: SUPPLY2 (schematic - SUPPLY2.GIF) This supply provides a variable but regulated DC voltage from about 80 to about 180V DC at currents up to at most 50 mA. The regulation source is neon bulbs. There is a 6AQ5 "cathode follower" driven from a pot connected to the neons. The output ripple from this supply is less than about 3 mV, and it needs no choke. The key to providing this kind of performance is the use of a tetrode (or pentode) "pass" tube, and filtering the screen grid (G2) source. Then, the relatively high dynamic plate resistance makes the ripple on the plate relatively unimportant, so long as the tube is kept in negative grid region. In addition, the plate voltage can drop below the screen, and still maintain regulation. The 470k / 0.1u capacitor on G1 filters any noise from the neons and the pot. There is almost -no- ripple at this point. The neons are not critical as I tested this with about 5 different "types" of NE-2 style neons. You can use one of the neons as a front panel "indicator" if you like. One caution.... some neon bulbs come with a resistor built in, so they can be connected directly across the mains. Won't work in this circuit. The output impedance is approximately 1/gm of the 6AQ5. A 6V6 will work exactly the same in this circuit. It should work even better with 6Y6 or 6W6 due to their higher gm. There's no reason to use the higher heater and plate capabilities of these devices, though. Notice that the ripple reduction is more related to the effective "gain" of the circuit. The heater supply provides a regulated 6.2VDC at about 0.6A. (yes, you can run 0.9A for a while to test 5687's). It uses an old LM309 regulator, with 2 diodes lifting the "ground" pin to raise the regulated voltage to 6.2V or so. This device was picked over the LM317 adjustable regulator as it has a lower minimum input to output voltage requirement to maintain regulation. This allows the 6.3V winding from the transformer, rectified by Schottky diodes, to provide regulated 6.2V. On my unit, I used a 500k linear taper pot for the voltage control, and as it had a switch on it, it also provides a power on/off function. By the way, this circuit makes a dandy preamp power supply. Calibrate the plate supply voltage by placing the meter to the ANODE range (after calibrating the meter - see below). NO TUBE in the sockets of the noise instrument. Place convenient "ticks" at positions of the pot, for example at 80, 100, 120, 140, 160, 180 volts. Parts list is located later in this document. NOISE INSTRUMENTATION (schematic - NOISEP1.GIF) There are 5 sections to the noise instrumentation. Each section is separately described. I. Additional Power Supply Section: For the HT voltage, an additional LC filter is used to take the <3 mV ripple to about 2 microvolts. This provides adequately "pure" DC so that it does not influence the noise readings of tubes with relatively high plate resistance, like the 12AX7. The heaters of the tubes to be tested are fed with a big "rheostat" so that the heater voltage can be adjusted, AND heater current can be derived (useful for the 6DJ8/6922 discernment). The rheostat is "calibrated" with an ohm meter, put tick marks at 0, 1 ohm, 2 ohm etc. The 0 ohm should be full clockwise, 10 ohm full counterclockwise. There is a Maxim ICL7662 IC (U4) that is used to generate -6V from +6V. The op-amps in this instrument run generally from +/- 6 volts. The meter amplifier uses the "raw" voltage, it is filtered using 220 ohm/100uf for the log amp section, and filtered again with 1k/470uF for the preamp. By the way, the Maxim part is quite useful for generating a reasonable negative voltage. It will provide 10 or so mA of the negative of what you put in, from 4.5 to about 20 volts. (That is, if you put in +15 volts, it will provide -15V output). This is useful for generating fixed bias sources. There is another example of its use in the BATTAMP schematic on these web pages. II. Metering Circuit This uses a low power op-amp as a meter amplifier. There are 3 basic configurations. One is a gain of 0.5 with 1M input impedance (10 volts in produces -5 volts output), one with a gain of 0.025 with 2M input impedance (200 volts produces -5 volts output) and one with a gain of 4.7 (1 volt produces -4.7 volts output). The meter is driven by a series "calibration" resistor (R39), which then sets the scale. This part of the circuit is adjusted by setting the meter switch to the heater setting, with no tubes in the test socket. Adjust R39 for 0.62 on the meter scale (6.2V). This caibration should be the first "adjustment", before the anode voltage pot is calibrated, or the noise "zero" is set. The op amp is "integrated" with a 0.1 uF (C16) primarily to "average" the noise in the noise position. If you like, you can paint the scales shown in the meter calibration above on the meter. This helps you using the instrument; you don't have to remember the scaling. "Whiteout" or similar correction fluid works well. The meter I used was a $1.00 special with wierd calibrations on it, so I just painted over the existing calibration and wrote my own. (You can see this in the picture of the unit: NOISE1.GIF). By the way, almost any standard OP-AMP can be used here, as long as its maximum negative output voltage is greater than -5 volts with +/- 6 volts on it, and drives at least 1 mA. III. Socketting and Tube Setup Fairly straightforward. I chose to add 2 socket configuration switches (S1 and S2) to select between 9A and 9AJ (moves heater leads) and between 9AD and 9CQ (moves grids leads). This eliminates 2 sockets, and cuts the wiring down a little, which is important in noise testing. Keep the plate connections short as possible, as this is going to affect the residual noise of the instrument. Notice that there are 5 capacitors in the cathode path. These are placed physically as alluded to on the schematic, namely, near pin 6 of V2, pins 3 and 8 of V3, pin 3 of V4 and near the switch S20. R35 and C19 should be located near the tube sockets or S3, with a wire running from this junction to the metering switch S21. This is intended to again keep the plate lead noise down. I have found no stability problems with my unit, but it is always a good idea to keep the plate and cathode leads separated. Notice that the 2 testable pentodes are tested in triode mode, with their screen grids connected to the plates. If you operate them in this mode, the measurement you get with this instrument will be accurate. If you operate them in true pentode mode, the noise you will obtain will be 10-20% higher, due to the partition noise (random division of electrons to the plate or to the screen). IV. 60 dB Preamp There is a 60 dB preamp made from a low power op-amp and a single low noise transistor. I chose this for a lot of reasons.... the op-amp is easily available (Tech America, Mouser etc) and cheap. You can use a low noise LT part here, but it's slightly harder to get and cost more, and draws more power, so it's harder to "filter" the supply voltage. The circuit produces about 60 dB gain with a bandwidth set to about 20 kHz, by C11 and C2. Because of the really low levels and high gain, take care in building this portion. I located it near the tube sockets and put it in a copper/ mu-metal shield. This can be seen in the picture NOISE3.GIF on these web pages. The output can then be run via shielded cable to the log amp. There is also an equalization switch that selects between flat and "RIAA" equalized noise. The flat position is simply a 10dB attenuator. The equalized position provides 10 dB boost for low frequencies and 10 dB cut for high frequencies, to simulate the effects of RIAA equalization. No attempt was made to put the "correct" RIAA network in the instrument, but this switch does provide a useful indication of the performance of the tube under test in either a flat configuration (mic amps) or in a phono preamp. There DEFINITELY is a difference in performance, either way. V. Log Amp / Rectifier This unit uses a log converter to gather RMS noise measurement info. The noise is log converted, using the exponential characteristics of 2 back to back diodes. U2A (a quad op amp - TL064: Nationals LF series works just as well) provides a buffer and drives a twin T 60 Hz notch filter that drives the AC log amp U2B. The signal here is roughly 1V pk-pk of a very rounded square wave (it's log remember) when looked at with a scope. (On pin 7 of the TL064). This signal is "precision rectified" by U2C. This converts the log AC into DC current. The op-amp "cancels" the diodes thresholds, providing a substantially linear rectification over the range of voltages expected (0.3V to 1.2V). You will notice a slight "curvature" in the characteristics shown for the meter calibration (above). The pulsating rectified DC is smoothed by integrator U2D. The overall gain is set so that 300 microvolts of noise at the plate of the tube being measured produces full scale on the meter. (Slightly over one volt DC at U2 pin 14). If you want to read higher noise voltages, you can lower the value of R51. For example, if you change R51 from 66.5k to 51.1k, you noise "meter scale" would be: 0 <2 uV .1 <2 uV .2 3.2 uV .3 8 uV .4 20 uV .5 43 uV .6 95 uV .7 190 uV .8 360 uV .9 610 uV 1 1000 uV To calibrate this portion of the instrument, set the meter switch to the NOISE position, and ground TP1 (pin 3 of U2). Adjust R37 for zero on the meter. INSTRUMENT OPERATION: Plug in your tube to be tested, and set up the desired operating conditions by setting up the plate supply voltage, the plate load, and the cathode bias resistor. (Use the settings you intend to operate at, or set per the tube manual). You can alternatively set the plate load and cathode bias resistors to obtain slightly over half the plate supply voltage at the ANODE VOLTAGE meter position with perhaps 1 volt on the cathode, as shown on the CATHODE voltage meter position. The heater rheostat should be fully clockwise. To measure the noise, switch to the NOISE position on the meter. Went full scale didn't it? Now... put a tube shield on the tube and wait for the meter to settle (takes a couple of seconds). The noise you read is the PLATE noise, to figure equivalent input noise, you'll have to divide this number by the expected circuit gain. Here's some rules of thumb if you don't know how to do this..... Tube type. Expected gain 12AX7 60 12AT7 40 12AY7 29 12AU7 12 12BH7 13 6SL7 40 6SN7 15 6CG7 15 6DJ8 25 417A/5842 35 5687 13 5879 13 EF86 14 Examples... 1. Your 12AX7 shows a noise of 60 microvolts. The equivalent input noise is 1 microvolt. 2. Your 12BH7 shows a noise of 52 microvolts. The equivalent input noise is 4 microvolts. 3. Your 5842 shows a noise of 30 microvolts. The equivalent input noise is 0.86 microvolts. Heater Current: Read the HEATER voltage with the heater rheostat fully clockwise. (Should be 6.2 volts.) Not change the rheostat to 2 ohms. and read the heater voltage. Subtract the 2 readings. Divide by 2 to obtain the heater current in amps. Examples: 1. Reading was 6.2V and 5.6V. Current is (6.2-5.6)/2 = 0.3A 2. Reading was 6.2V and 5.4V. Current is (6.2-5.4)/2 = 0.4A (Expected readings from 6922 and 6DJ8 respectively). Notice that as you increase the resistance, the voltage difference will increase, making current differences more apparent. Parts List - Noise Instrument Qty Reference Part ______________________________________________ 3 C1,C3,C9 470u 6V elec 1 C2 5 pF ceramic 5 C4,C5,C6,C7,C26 47u 15V tant 2 C8,C19 0.1u 200V 1 C10 2.2u monolythic 1 C11 200 pF SM or ceramic 1 C12 220u 15V elec 1 C13 2200pF SM, mylar etc. 4 C14,C17,C18,C24 0.1u 25V ceramic or monolythic 1 C15 10nF 1 C16 0.2u 1 C20 100u 350V elec 4 C21,C22,C23,C25 100u 10V tant 6 D1,D2,D3,D4,D5,D6 1N4148 (or 1N914) 1 D7 4.7V 250 or 400 mW zener 1 L1 6H 40 mA or greater 1 M1 0-1 mA meter 1 Q1 MPS-A18 2 R1,R34 499k 3 R2,R5,R47 1k 4 R3,R7,R8,R42 100k 1 R4 100 3 R6,R33,R40 1M 5 R9,R32,R43,R46,R50 10k 1 R10 32.4k 1 R11 130k 1 R12 68k 1/2W 1 R13 33k 1/2W 1 R14 16k 1W 1 R15 8.2k 2W 1 R16 4k 3W 1 R17 2k 3W 2 R19,R18 1k 5W 2 R20,R22 26.7k 1 R21 13.3k 1 R23 1.3k 1 R24 649 1 R25 332 1 R26 162 1 R27 80.6 1 R28 40.2 1 R29 20 2 R31,R30 10 1 R35 49.9k 1 R36 4.7M 1 R37 100k trimmer pot 3 R38,R44,R49 20k 1 R39 5k trimmer pot 1 R41 470k 1/2W 1 R45 10 10W pot 2 R48,R52 220 1 R51 65.5k 3 S1,3,12 SPDT mini toggle switch 1 S2 DPDT mini toggle switch 16 S4-11,13-20 SPST mini toggle switch 1 S21 2 pole 4 position rotary switch 2 U1,U3 TLC271 1 U2 TL064 1 U4 ICL7662 1 V1 Octal Socket 4 V2-V5 9 Pin Socket 1 -- Chassis and cabinet 3 -- knobs Notes: All resistors are metal film or wire wound (power). Supply 2 Parts List Qty Reference Part ______________________________________________ 4 C1,C3,C4,C7 100u 350v elec 2 C2,C6 0.01u 1kV ceramic 1 C5 0.02u 300V 5 C8,C11,C12,C14,C15 0.1u 25V ceramic or monolythic 1 C9 0.1u 200V 1 C10 100u 10V tant 1 C13 10,000u 15v elec 2 D2,D1 1N4007 2 D3,D5 1N4004 (can use 1N4007's here too) 1 D4 3A 35V Shottky diode bridge 1 F1 1A fuse and fuseholder 1 J1 Power cord 3 LP1,LP2,LP3 NE-2 1 R1 470 1W 1 R2 47 5W 1 R3 10k 1W 1 R4 390k 1W 1 R5 330k 1/2W 1 R6 47k 1/2W 1 R7 500k pot, linear taper 1 R8 470k 1 S1 SW SPST (may be part of R7) 1 T1 secondary: 230-0-230 50mA, 6.3V 2.5A 1 U1 LM309K 1 V1 6AQ5 tube and 7 pin socket