The schematic attached is two stages of amplification followed by a DC coupled cathode follower that I found in an article. It's intended to swing a lot of voltage with low output impedance to drive a demanding triode. It seems to me to be overly complicated, but that just might be because I don't follow the designers intent.
I can't see the purpose in elevating the cathode resistors in the first and second stage with the 2k7 resistors. It requires a cap on the V1 grid that would best be avoided IMHO.
I also can't see the purpose in dropping the B+ with the 3K3 resistor and the 47uf cap. It seems needless since the 6SN7 plate voltage will by necessity remain at 205v with or without.
Perhaps the goal was to hit the 205 plate voltage necessary for dc coupling to the cathode follower, but that same operating point can be hit without these elements. A 63K plate load will get to the same operating point with a 2.2 cathode resistor at the 430 B+.
I'd assume that the heater voltages are elevated although that's not mentioned.
I bread-boarded this circuit without these elements and it performed quite well on the bench. I also replaced the plate resistors with a CCS and it showed quite an improvement in swing and a better wave form. But still, I remain puzzled.
Can anyone see a need for the the lowered B+ or the elevated cathodes?
I can't see the purpose in elevating the cathode resistors in the first and second stage with the 2k7 resistors. It requires a cap on the V1 grid that would best be avoided IMHO.
I also can't see the purpose in dropping the B+ with the 3K3 resistor and the 47uf cap. It seems needless since the 6SN7 plate voltage will by necessity remain at 205v with or without.
Perhaps the goal was to hit the 205 plate voltage necessary for dc coupling to the cathode follower, but that same operating point can be hit without these elements. A 63K plate load will get to the same operating point with a 2.2 cathode resistor at the 430 B+.
I'd assume that the heater voltages are elevated although that's not mentioned.
I bread-boarded this circuit without these elements and it performed quite well on the bench. I also replaced the plate resistors with a CCS and it showed quite an improvement in swing and a better wave form. But still, I remain puzzled.
Can anyone see a need for the the lowered B+ or the elevated cathodes?
Yes, that must be it. I would think you mean heater elevated to 115 v, given the 215v cathode voltage in the follower and 16v cathode in the first two stages?
Last edited:
Yes, of course.
Can you offer an explanation why the first and second stages are not decoupled from one another?
This is an area don't understand particularly well. Oscillation is the issue I believe, and gain between stages is an aggravating factor and PSRR is a mitigating factor. If I'm correct about this, then wouldn't the resistor and the reservoir cap be placed between the first and second stages of amplification (where there is high gain) rather than before the second stage and cathode followers (where there is none)?
Sets the gain. Lowers THD.
If "HT" is not PERFECTLY stable (it never is), then the big chunks of current in it will sneak-back to the input stage plate feed. Say V3V4 current increases to make a positive peak output. HT will fall. If this sneaks-back to V1 plate circuit, V1 plate will fall some fraction of that. The fall at V2 grid will drive V2 plate high, with gain. This will drive V3V4 grids high, making *more* current flow. Positive feedback.
If forward gain (V1 plate to V3V4 load on HT's impedance) exceeds losses in the B+ rail, it will oscillate. Because B+ is loaded with caps, it won't sing high, it sings low where the cap impedance is high and does not reduce sneak-back. "Motorboating".
Since you do not observe this, you may have a very solid "HT". I have seen 7 or 7 cascade stages on one B+, loaded with 470uFd and an emitter follower (<1 Ohm). Also there may be some synergy here which makes HT bounce not worst-case.
As a general no-brain-pain guide: every two cascaded stages should have their own R-C decoupling network with a time constant far below the bass limit of the forward gain.
Last edited:
Jones asserts in his book that if the two stages are out of phase and the current is identical (3.55mA in this example), then decoupling is not required.
The resting currents are the same but working signal voltages aren't so some feedback would occur through the power rail.
I have found it best to decouple every stage.
I’d like to read that in full. Third edition? If you can offer a chapter or page number I’d be grateful.
Thanks for the suggestions. I can understand the need for stoppers. I’ll add them to the drawing.
And you bring up the heater elevation. I’m planning on 165v to satisfy the cathode followers. The cathode followers swing a bit less than 150 volt peak so let’s say 365 to 55. I need to have the heaters at a minimum of 165v to satisfy the 200v DC+peak heater negative with respect to cathode. It’s within the heater negative with respect to cathode of 200v dc plus signal for the first stage.
Current equality is not needed. Since the signal levels are different, the signal *currents* are never equal, unless you "cleverly" design for that goal.
However the amplifier in *this* thread is three-stage. That 3rd stage plate-side is in-phase with 1st stage plate, and sneak-back will be positive feedback. The hasty-hack rule is "two stages on one decoupler". That rarely leads to trouble (if the decoupler is effective beyond where amplifier bass gain falls off). It can sometimes be violated, and indeed a $500 regulator on a $5 amplifier may allow a lot of cascaded stages on one HT.
I built up the circuit again today with the changes everyone contributed as well as adding the ccs plate load on the second stage, and then tested the circuit.
The most important issue I found that I could no longer make the required swing from the second stage after decoupling that stage from the B+. I lost about a volt peak for every volt reduction in B+ (no surprise there really).
Below I have redrawn the schematic with a new plan. I'm separating the first and second stages B+ so they are in parallel rather than series and reducing the current in the second stage. This of course greatly reduces the voltage drop across the decoupling resistor. At 2ma with the CCS I should regain virtually all the swing due to the sharpening curve of the zero bias line at low currents.
I'll test this circuit on the bench tomorrow and see how it does.
The most important issue I found that I could no longer make the required swing from the second stage after decoupling that stage from the B+. I lost about a volt peak for every volt reduction in B+ (no surprise there really).
Below I have redrawn the schematic with a new plan. I'm separating the first and second stages B+ so they are in parallel rather than series and reducing the current in the second stage. This of course greatly reduces the voltage drop across the decoupling resistor. At 2ma with the CCS I should regain virtually all the swing due to the sharpening curve of the zero bias line at low currents.
I'll test this circuit on the bench tomorrow and see how it does.
I'm looking for a bit over 300vpp. The output load is the 220K load resistor and the Miller capacitance of an 845.
It always seems a bit easier to DC couple with the same type and the same OP. I'll will have a look at those curves though and see how it looks.
Last edited:
If a few-Volt drop of B+ causes the design to fail spec, you are skating too close to the edge. Everything drifts. It may work when new and "fail" (however marginally) a month later.
Of course clean speech/music will be <312Vpp 99.9% of the time, so a drift to say 306Vpp may be unnoticed.
The motorboating is from stage 3 to stage 1. Stage 2 can be thrown-in with either side. Feed it the same as stage 3.
Your "heater bias" needs a low AC impedance to absorb H-K hum leakage. Cap it.
If you truly expect 312VPP at those cathodes, I fear you are pushing the insulation. I see that golden-age 6SN7 had high H-K ratings; I wonder if all did, including modern production. The safe path is a separate heater transformer for the CFs, for each channel. The downside is another hundreds-pFd of capacitance.
What are you doing that needs cathode followers? A high-current plate-loaded stage is a beefy driver. With a CCS the up-swing can be quite large.
Of course clean speech/music will be <312Vpp 99.9% of the time, so a drift to say 306Vpp may be unnoticed.
The motorboating is from stage 3 to stage 1. Stage 2 can be thrown-in with either side. Feed it the same as stage 3.
Your "heater bias" needs a low AC impedance to absorb H-K hum leakage. Cap it.
If you truly expect 312VPP at those cathodes, I fear you are pushing the insulation. I see that golden-age 6SN7 had high H-K ratings; I wonder if all did, including modern production. The safe path is a separate heater transformer for the CFs, for each channel. The downside is another hundreds-pFd of capacitance.
What are you doing that needs cathode followers? A high-current plate-loaded stage is a beefy driver. With a CCS the up-swing can be quite large.
In 1994, as was the fashion, I built an 845 amplifier and used paralleled halves of type 6SN7 for drivers, plate loaded and cathode resistors un-bypassed, from a +700 Volt DC supply. Despite the precariously high impedance at the 845's grid, frequency response was primarily limited by the output transformers, even the fancy Audio Note's. You may not need or want cathode followers; just food for thought.
If I were to make another one, I'd use triode connected 6BQ5 / EL84 as drivers, again from about a +700 Volt supply. Look at Telefunken's triode curves and you'll see what I mean.
All good fortune,
Chris
If I were to make another one, I'd use triode connected 6BQ5 / EL84 as drivers, again from about a +700 Volt supply. Look at Telefunken's triode curves and you'll see what I mean.
All good fortune,
Chris
Sorry, but one more piece of un-asked-for advice: If you feel yourself becoming a fashion victim (like me) and start building the thing with milsurp iron and oil-filled power supply caps, don't - do not - build it all on one chassis. Two channels and their all-in power supply seems fine now, but someday you'll be old and can no longer pick it up by yourself.
It's more embarassing than you imagine now. All the best fortune,
Chris
It's more embarassing than you imagine now. All the best fortune,
Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.