|
K6JRF's Page formerly W6FZC AL-572 DBC and AL-1500 EBS1 Ckt Analysis |
|
This page details some improvements that can be made to the Ameritron AL-572 Dynanmic Bias Ckt (DBC) and the AL-1500
Electronic Bias (EBS) Ckts. By use of 5Spice analysis, the ckt operation is detailed and recommendations are made. I also switched from MSim's "PSpice" to a newer, more compact analysis program called "5Spice" written by Andresen Software. It features almost intuitive operation, has many powerful features and and allows the import of various Spice models so that the ckt can be simulated using the actual real world parts. Click here to see more information on this fine circuit analysis product. |
|
Received this email . . . Hi OM, Your model is all messed up! The analysis is meaningless because you applied audio frequencies to the input instead of radio frequencies. The lowest possible frequency on the input is 1.8 MHz, not 10 Hz or even 3,000 Hz. Why would anyone model an RF detector with audio frequencies applied? 73, Tom W8JI His statements took me back a bit. One should know that a "modulation" envelope is what SSB produces (not a 'carrier'), so in the limit (using math jargon), a 50hz 'tone' modulating the SSB Tx at some RF frequency is one example. Since low frequencies (50hz) are easily produced in the newer radios (FT-2000, FT-9000, etc) the problem centers on the inability of most EBS ckts to pass these low frequencies @ low level b/c of two (2) main factors: improper time constant (too short) and input passband skew (shifted toward high frequencies). Perhaps W8JI has not seen the comments from Richard, AG6K on RF actuated EBC ckts; "This is a desirable departure from the RF-actuated electronic cathode-bias switch circuits that have appeared in ham magazines. At first, "RF-actuated" sounds like it might be wonderful. However, these "RF-actuated" circuits result in the cathode bias-voltage being rapidly switched between nonlinear-cutoff and linear operation while the RF relays are in transmit. This often causes transmit audio to sound rough on softly spoken syllables and increases the IMD products that the amplifier generates." As he correctly stated, the analysis must be taken far enough to INCLUDE lower passband (50hz - 4Khz at least) so that the tube's bias is NOT rapidly changed or switched off b/c of incorrect time constants and low ckt gain in the for low frequencies used in the present EBS ckt design. Prior to my analysis (AL-572, AL-1200, AL-1500, etc), the EBS ckt design only considered a restricted passband as evidenced by the present EBS ckt components in Ameritron amps. I'm surprised that he was not able to see (and hear!) the problem. My ckt analysis simulates exactly how the ckt operates. BTW, it's impossible to be wrong using SPICE(!) b/c there's no assumed "MODEL". The complete ckt is captured and simulated using Spice's "transient" and "steady state" analysis tools. Correct EBS operation can't be seen by just assuming that only an RF carrier is passed. That leads one to W8JI's incorrect conclusions. The unmodified EBS1 ckt will marginally pass low frequency of 400hz. Modified w/ my recommended parts, it passes 50hz; "passing" means that the tube's bias will NOT be switched off for any audio frequency in the SSB passband. For "CW" operation, W8JI is correct, the ckt works nicely as is! And at least one independent "proof" was demonstrated by Tom, K6NHK on his AL-800H. He exhanged the recommended parts and the low frequency distortion was removed. His SSB Tx audio is perfect. I've received email from other Hams that had the same results. Of course, my AL-1500 EBS1 ckt was the original source of the problem and it too is now working perfectly. |
|
Introduction: 
The Ameritron AL-572 Amplifier runs four (4) 572B tubes that produces 1200W CW. It has many features but one of the best
is its "Dynamic Bias Ckt" [DBC] that eliminates hundreds of watts of unnecessary heat generation
in the power amplifier tubes. The result is cooler operation and longer component life.Problem: As I've stated on the "Vox Mod" page, I really enjoy VOX operation but it's a bit 'hard' on my amplifier in that it forces the DBC to continually cycle. There are no "moving-parts" per se but the strain on the switching transistor makes it suseptibale to a high voltage transient which can exceed the collector-to-emitter breakdown voltage [Vceo] causing it suddenly "die"! Then the circuit reverts to a continous bias mode so the static plate dissipation now takes over. About 450+ watts are wasted in heat and that, of course, raises the overall temperature level and, in some cases, may contribute to glowing red plates under a long SSB transmission. As you see in the comparison table, the high voltage switching transistor, Q103 has a 'breakdown' voltage, Vceo = 60V and is very suspectible to catastrophic failure for no apparent reason. When you transmit again (key down, no audio), you see quesient plate current with no RF drive signal, that's a sure sign that the transistor has failed. The table compares the parameters of the transistor. DBC Operation: Conventional bias circuits force high power linear amplifiers to dissipate hundreds of watts during low or no signal periods. This creates needless heat, since virtually no dissipation is required unless the amplifier is being driven with large signal levels. The AL-572 contains an exclusive bias circuit that reduces the idling (quiescent) current very close to the tube's cut-off region. The AL-572 power amplifier tubes have a full resting period of very low dissipation between dots and dashes on CW and between words on SSB. The lower idling current reduces component temperature on both CW and SSB. If only a few milliwatts of RF power are applied to the amplifier, the quiescent current will increase. Linearity remains excellent with this circuit because the tube can remain biased for class AB operation without unnecessary standing dissipation. The best of both worlds. The DBC is located on the Power Supply Board. Diodes D101 and D102 rectify a small sample of the RF drive voltage. This voltage is applied to the base of dc switching transistor, Q101. If Q101's base is driven with a few milleamperes of current from the RF sampling circuit, Q101's collector will pull the base of PNP transistor Q102 low. This turns the dc switching transistor, Q102 on. When Q102 is on, zener diode, D103 (9.1V zener voltage) is connected between the collector and base of Q103. D103 sets the operating bias. This zener applies forward bias to Q103's base whenever the collector voltage of Q103 exceeds the voltage of D103. This forward bias will turn Q103 on harder and reduces the collector voltage. If the collector voltage is less than the breakdown voltage of D103, Q103 will move towards cut-off and the collector voltage will increase. Q103 functions as a current buffer for zener diode D103. Check out the 5Spice transient analysis charts below. An improved part for this ckt is the "TIP122" darlington transistor. It features Vceo of 100V giving an added 40V safety margin. The second advantage is the "darlington" configuration is the added current gain meaning that less drive is needed to saturate Q103.
5Spice Analysis: By capturing the schematic into 5Spice, it's possible to analyze the critical parameters and see what can be improved. The circuits operation is described in the 'DBC Operation' section above.   The chart shows the AC Frequency response of the DBC. As noted in the opening paragraph, the ckt must pass all of the SSB Tx audio frequencies w/ attenuation. This means that all frequencies in the SSB wavefront should pass equally and it should delay in turning off so that so any audio "effects", if any, are fully "passed". As you can see, the BLUE line in the chart shows that the DBC's passband is very restricted; 100hz is at least -21dB down. Obviously not desirable! The RED line shows the proper response when C102 is changed to 0.047uF. The Spice ckt simulation was run to map the ckt's transient performance and it produced the chart below.   The TIP122 has much more current gain (hfe) and so the resistor, "Rswamp" reduces some the current gain to prevent electronic "chatter" in the DBC. These two (2) parts will dramatically improve the DBC's operation and reliability of the Ameritron amplifier. In over 4 years of use, there's not been even one failure in the DBC. I highly recommend these changes. Parts Access: Capacitor, C1 and C10 can be soldered in parallel to the existing caps. Transistor, Q103 is located on the wall divider between the power supply section and the tube compartment. The TIP122 is the same size (TO220) as the MJE3055 so the replacement is easy. The resistor is an "add-on" part that mounts on the ends of R107 and D103. A 15 min modification and nothing needs to be removed. All parts are on the Power Supply Board, PN 50-0080-1. Detailed Parts Info: These are common parts and can be found in most electronic stores. The best pricing can be had from Mouser Electronics: Ref Des ------ Value ------- Price/Ea ------- Link -------- Remarks C1 --------- 0.047uF ------- $0.57 ------- -here- ------- 600V rating is ok: 50W drv = 141Vp-p! C10 --------- 10uF --------- $0.08 ------- -here- ------- 50V rating! Q103 ------- TIP122 -------- $0.52 ------- -here- ------- 100 Vceo + high "beta"! Rswamp ------ 47 ohm ------- $0.33 ------- -here- -------- 47 ohm, 1% film resistor! Comments: The Dynamic Bias Ckt (DBC) transistor + resistor mod described has been in use for 4 years and have proven to be reliable and has withstood all kinds of operator "mis-operation"! The original parts failed continually so this mod is a very reliable, rugged solution to the problem. And it's cheap! The capacitor mods make the ckt respond to the modulation frequencies and eliminate the distortion of the original ckt configuration. Electronic BiaS Ckt [EBS]: 
As you saw in the AL572 analysis above, the AL-1500 linear also supports an a "electronic bias ckt" which is called the
EBS (Electronic BiaS) ckt. To me, all linears should have one! It saves the useless waste of "idling" power of
480 watts (3850V @ 125ma)! That said, recently, I found that this ckt restricts SSB Tx audio b/c of circuit-time-constants. This section is devoted to analyzing these problems and has suggested ckt improvements. It's mounted near the opening of the fan plenum area and has a small "detector" board mounted on a standoff, just above it. As you see, I've marked the "missing" components which makes a difference in the simulation. EBS Schematic: The schematic was copied from the user's manual and translated into "5Spice" so it's not the original but it's much easier version to read. It similar to the AL-572 EBS ckt and generally operates in the same manner.  The EBS-1 features a remote detector board that uses a pair of Schottky diodes, driving a bipolar transistor switch (Q5). Q5 is a fast-attack, slow-decay switch with C10 setting the decay time and R4 limiting the charging current. Delay or “hang-time” is an additional period when bias remains at normal operating levels after drive power goes to zero. The EBS-1 remote detector board serves as an "RF" high impedance sensor. It's mounted near the source of RF, preventing the EBS-1 ckt from having minimal effect on input SWR. The switching and control board is mopunted directly below the RF detector board. To be a 'good' ckt, it should switch on quickly so that the SSB Tx wavefront is fully allowed to pass. This implies that all frequencies within the wavefront should pass equally and it should delay in turning off so that so any audio "effects", if any, are fully "passed". 
Another problem was found when doing the 5Spice AC simulation; transistor Q5 (MPSA05) does NOT have enough current gain (BETA)
to be the key "switching" element in the ckt. That transistor determines whether the signal is "valid" by switching from "off"
to "on". A low-gain transistor further skews that ckt's repsonse to the high-frequency side. This is evident for a low level
SSB drive of 10 watts! The new transistor, MPSA29, is basically a "darlington" transistor meaning that there are two transistors inside the part feeding each other. The current gain (BETA) is extremely large (typ 1500 to 10,000) and that translates into "high sensitivity" which is a required for proper operation of this ckt. This part is similar to the TIP122 used in the AL-572 DBC ckt and was subsituted for the MPSA05 (hfe = 100) which was very short on required gain! The chart above shows the frequency response of the EBS1 ckt. The BLUE [C1= 0.001uF] line shows that it's -28dB down at 50hz, so its response is skewed to the hi-frequency side. This causes a severe loss of lo-frequency audio. The RED line shows the response centered within the passband allowing all SBB audio (from 20hz to 5Khz) to pass w/o attenuation. Both plots reflect the replacement of Q5 with a "MPSA29" transistor. Other parts such as 2N7052 or 2N7053 are compatible. 
In addition to a balanced freq response, it must turn on and off faster than the rise time of voice waveforms. This is not a
problem since the choice of the R-C time constant is easily chosen. The problem in this ckt (and in ALL half-wave rectifier
ckts), is the "hole" left by the "dead" time of frequency in question. For 50hz, it's 10msecs. In other words, the time
'delay-time' of the ckt must account for this so that it doesn't start to "turn-off". That's what happens using the "stock"
Rs and Cs. The chart show the repsonse to a SSB signal envelope consisting of 50hz audio component at 10watt drive level into the stock EBS1 ckt. This, for my AL1500, produces apx 650 watts output. As the chart clearly shows, the ckt NEVER FULLY SWITCHES ON! After 210msec the ckt oscillates around 4V to 9V level making the 8877's bias fluctuate from 16V to 21V. It's clear that stock EBS1 ckt is never quite fully "ON". At 300msec, it continues to stay partially "on". For a SSB signal consisting of "1000hz", the ckt is fully "ON" and allows that to pass w/o attenuation. But for a 50hz SSB signal, there is severe distortion b/c the 8877 tube is varying between cutoff and partially-on bias. In the AL-1500 amp, the stock value for C10 is 0.33 µF and this results in apx 150mS of bias hang time as per 5Spice analysis depending on the input SSB drive level. To increase the bias hang time, C10 can have a range from 1uF to 10 µF. Since 250ms to 350ms would be more desirable, C10 should be increased. And indeed, a larger value is needed to ensure that the ckt doesn't attempt to turn off for low frequency SSB envelope signal consisting of 50hz. 
The values for C1 and C10 are critical of ESSB b/c they determine the audio passband response. A serious attenuation of
frequencies below 500hz occurs w/ the "stock" values for C1 (0.001uF). And the "half-wave-hole" requires a much larger value to serve as
an "integrator" over the time that ensures that the ckt stays "turned-on". This is important for ESSB radios and the changes should be
made if you wish to properly pass low frequency ESSB Tx audio. The chart shows the EBS-1 ckts response to SSB envelope consisting of 50hz with the C1, C10 and Q5 value changes. Now the ckt is fully turned on in less than 2msec and easily passes the 50hz ESSB Tx audio. There is some passband ripple but not enough to effect the class AB bias of the 8877 tube. 
This chart shows the 5Spice simulation analysis with the component value changes for C1=0.047uF , C10=1uF
and Q5=MPSA029 using a single trigger "pulse".
These new values eliminate the low frequency response cut off and these values guarantee that the ckt will pass all low frequencies without
attenuation or distortion.The charts show the reponse of the EBS-1 ckt for a SSB single "pulse" signal. The "turn-on" time (green line) is a bit slower (apx 400usecs) than before but is fast enough to ensure that there is no attenuation of the low frequency signal. 
The second chart shows the same response except over a large time frame so the ckt's response to a single input trigger can be analysed.
The "hang-time" (green line) is apx 350msecs but by 200msecs, the tube is 'cutoff'. It's slower than the stock (a good thing) b/c of the
larger value for C10 (1uF) and allows any audio-effects such as reverb, to pass thru the linear w/o distortion or being cutoff.
If you want to shorten this time, use a 0.47uF cap for C10. The hang-time will be about 120msecs. Q1 is an MJE3055 transistor which has 60V collector to emitter breakdown (Vceo) voltage. I MAY replace it with a TIP122 with a Vceo of 100V. This was the main replacement in the AL-572 amplifier's bias ckt as you read in the first section. Time will tell if the present transistor can continue to perform without failure. So far after 6 months operation of the linear, transistor (Q1, MJE3055) has not failed. Final Schematic: The schematic below has all of the mods incorporated. 
Parts Info: Some of these parts are very common and can be found in most electronic stores such as Radio Shack. However, much better prices can be obtained from Mouser Electronics and that's what I've used. Click the link for each part to see Mouser's price and availability. Ref Des ------ Value ------- Price/Ea ------- Link -------- Remarks C1 --------- 0.047uF ------- $0.57 ------- -here- ------- 600V rating is ok: 50W drv = 141Vp-p! C10 --------- 1uF ---------- $0.08 ------- -here- ------- 50V rating! Q5 --------- MPSA29 ------- $0.18 ------- -here- ------- drop-in replacement; beta=10K! R8 --------- 22K ohm ------- $0.05 ------- -here- ------- 22K ohm, 1% film resistor C1 and C10 should be removed and replaced with the new values. R8 is soldered to the back side of the board so it should be replaced with the new value (22K ohm) resistor. Q5 must be unsoldered and replaced with the new part, MPSA29, which is a pin-for-pin replacement transistor. You can use an alternate such as a "TIP110" but the pins require some bending to fit the existing PCB pattern; I don't recommend it. Comments: Today (7/17/09), all of the EBS-1 mods were installed as shown in the schematic. Glad to say, that it's works as advertised! The lows are now passed as they should be. Very pleased and recommend the ckt mods. Reports for other ESSBers say the low frequency audio is clean and clear. Note that I've used "audio" terms and concentrated on the low frequency components contained in the SSB envelope. Certainly these audio signals are modulated via the process of making "SSB" signal at any RF center frequency. What many fail to understand is that low frequency audio (50hz for example) is normally much less amplitude than, say 1000hz, so with the low gain and skewed bandpass of most EBS ckts, these signals get "distorted" passing thru a stock EBS ckt unless the ckt has been optimized to pass such frequencies. The analysis here has clearly shown that the problem exists and the recommended changes do eliminate it! |
Send me  with your comments
|