jdrouin - please tell me you are not breadboarding an amplifier with high DC voltages in your living room using alligator clips to hook it all up. If you do not live alone, you really need to be careful with this hobby. Also, if you have pets, they could also brush HT and that would be devastating.
I thought we had this base covered already, but I see you needed to build it with a wimpy driver. Fine. Don't get the idea to bring it to a shoot-out once you put it in a proper box. Don't get an external sound card and software to do harmonic distortion measurements either. Also, I advise not to try and sell people on your design. Most of us who have designed (from scratch) and built more than a dozen amplifiers have "been there - done that".
If you want to something that will not distort immensely (3rd harmonic becoming quite dominant and ugly) after about 0.5W then you should consider some of the advice in the previous posts. There are many here on this board who provide very good advice based on much experience.
Sorry if I am not as carefully polite as 6A3sUMMER. I will attempt to say something nice: Yes, valve/tube amplification is a beautiful thing. 😉
Ian
I thought we had this base covered already, but I see you needed to build it with a wimpy driver. Fine. Don't get the idea to bring it to a shoot-out once you put it in a proper box. Don't get an external sound card and software to do harmonic distortion measurements either. Also, I advise not to try and sell people on your design. Most of us who have designed (from scratch) and built more than a dozen amplifiers have "been there - done that".
If you want to something that will not distort immensely (3rd harmonic becoming quite dominant and ugly) after about 0.5W then you should consider some of the advice in the previous posts. There are many here on this board who provide very good advice based on much experience.
Sorry if I am not as carefully polite as 6A3sUMMER. I will attempt to say something nice: Yes, valve/tube amplification is a beautiful thing. 😉
Ian
Last edited:
Member
Joined 2009
Paid Member
Jeez, somebody drank too much coffee.
Wimpy drivers - bring it on. I’m buying it. I already bought the same driver tube some years ago for just the same project. It’s been reported over and over again by designers going back a long time now, that this approach sounds darn good, if you know how to use it.
Wimpy drivers - bring it on. I’m buying it. I already bought the same driver tube some years ago for just the same project. It’s been reported over and over again by designers going back a long time now, that this approach sounds darn good, if you know how to use it.
Member
Joined 2009
Paid Member
https://www.diyaudio.com/community/...ivers-for-2a3-work-as-well-as-they-do.212673/
In this thread Soulmerchant says “Serious though.. it really shouldn't work but it does.” (Read the thread for the full quote, context etc.)
Let’s show some love for wimpy drivers 😎
In this thread Soulmerchant says “Serious though.. it really shouldn't work but it does.” (Read the thread for the full quote, context etc.)
Let’s show some love for wimpy drivers 😎
It seems a couple of people have become frustrated by the procedure I've been following, so let me just reiterate the intention I laid out at the beginning of this thread, which was to breadboard well-known published circuits, report on them (mainly for beginners, because of the lack of measurements available online), and use them as a reference point for making adaptations. In this case, I am moving on from capacitor-coupled circuits to try direct-coupled ones for the first time, so I started with the old Robin/Lipman and Garber circuits and will be adapting them to try other things, such as adjusting the operating points and bias to achieve better results. I wasn't trying to "sell" the adapted circuit posted just above, but rather to say 'here is an interesting point that others might find worth trying.' And, I like to write, and had a little moment of inspiration, and decided to put it a certain way because it's worth thinking about the meaning of what we do, too.
That being said, we had recently been discussing how to calculate the amount of current needed to avoid slewing in the output stage, etc., so I completely understand why this shift might have seemed like backpedaling. As I said, it was the starting point for a new phase of trying direct-coupled circuits, so perhaps I should have made that more clear.
Since we've established how and why a 2A3 needs 8-9mA from the input stage, I will next be swapping in a 5842 running at 150V/10mA. Should have an input sensitivity of about 1.5V. I might not get to it for a few more days, but attached is a screenshot of the LTSpice simulation. Will need to take into consideration that a LOT of heat will be dissipated at R3 and R6.
FYI, the breadboard is on a dedicated table in my small home office, out of reach from the dog. My kids are old enough to know not to touch it. The alligator clip is merely holding together three connections on the tab of a ceramic wirewound adjustable resistor because I need to manipulate it frequently, else I would have soldered them. I am always extremely careful around it, do not work on it when tired, etc.
Attachments
That low current drivers like 6SL7 and 6C6 not just work, but actually sound good is a surprise. My 2c: slew rate is important at higher signal levels, but at 0.1 - 0.2W output, the actual power needed to drive efficient speakers, a 0.8 mA driver will not limit slew.
Another important consideration is driver's plate voltage, which is the key factor in driver distortion. My rule of thumb is that driver's output voltage swing should not exceed about 20% of driver's Ua. So, for 50V peak, driver's Ua should not be considerably lower than 250V.
RC-coupling and DC coupling put serious constraints on driver's Ua. LC and transformer coupling don't.
Another important consideration is driver's plate voltage, which is the key factor in driver distortion. My rule of thumb is that driver's output voltage swing should not exceed about 20% of driver's Ua. So, for 50V peak, driver's Ua should not be considerably lower than 250V.
RC-coupling and DC coupling put serious constraints on driver's Ua. LC and transformer coupling don't.
Agreed. It's weird. 6SL7 SRPP (only 0.7mA) cap-coupled to 2A3 might still be my favorite circuit evaluated here so far. The DC 6SF5 can fall apart on some of the more complex or challenging music material, in ways that the SRPP doesn't, but it has a more capacious sense of space and is generally more relaxed and natural.
I also found, contrary to the wisdom a couple of you have posted recently, that increasing the signal drive in the 6SF5 DC circuit resulted in much better performance. The other night I listened to Kenny Burrell's Guitar Forms after the amp warmed up for an hour and decided to crank up the volume quite a bit, and boy did the orchestra sound great. But then I played Miles Davis' Sketches of Spain, which is similarly composed and orchestrated, and it really crapped the bed. But then I put on Jimmy Smith's Root Down and the bass was super authoritative and tuneful and the performance space, instrument separation, etc., were spot on.
The 6SL7 SRPP is more forward and bright sounding (not irritably so), but performs consistently well across music types. Actually, I think I'll try 6SL7 in simple and then SRPP topologies next, since I won't need to change much on the breadboard.
I also found, contrary to the wisdom a couple of you have posted recently, that increasing the signal drive in the 6SF5 DC circuit resulted in much better performance. The other night I listened to Kenny Burrell's Guitar Forms after the amp warmed up for an hour and decided to crank up the volume quite a bit, and boy did the orchestra sound great. But then I played Miles Davis' Sketches of Spain, which is similarly composed and orchestrated, and it really crapped the bed. But then I put on Jimmy Smith's Root Down and the bass was super authoritative and tuneful and the performance space, instrument separation, etc., were spot on.
The 6SL7 SRPP is more forward and bright sounding (not irritably so), but performs consistently well across music types. Actually, I think I'll try 6SL7 in simple and then SRPP topologies next, since I won't need to change much on the breadboard.
Last edited:
sser2,
Slew Rate considerations? True.
For example, we can test an amplifier for full power from 20Hz to 20kHz.
Testing is a great thing.
But the question arises:
When the recording has the Bass Viol putting out 5 Watts at 40Hz at the same time the piccolo also putting out 5 Watts at 4000Hz,
What happens? . . . OUCH!
If the [5 Watt] piccolo is causing slew rate distortion of the amplifier, then your hearing probably is in danger.
Is that Music?
Perhaps some can consider the above.
They might see why an amplifier that has slew rate distortion of high frequencies at full power output, Can Sound Wonderful on Real Music.
Purchase an entry level digital oscilloscope, set it for FFT from low frequency to 25kHz, and set the trace to Infinite persistence.
Now, play one of your favorite recordings all the way through. Notice how the FFT shows amplitude dropping off as the frequency increases.
Still not sure, then play another favorite recording and look some more.
If the high frequencies are not dropping off at high frequencies, then perhaps . . .
Your amplifier is clipping
The recording is clipping
It is Electronic Music (perhaps Not Good electronic music).
You need to consider the complete process from the music performer all the way through equipment to the end (your ears).
It is rightly called a System.
Slew Rate considerations? True.
For example, we can test an amplifier for full power from 20Hz to 20kHz.
Testing is a great thing.
But the question arises:
When the recording has the Bass Viol putting out 5 Watts at 40Hz at the same time the piccolo also putting out 5 Watts at 4000Hz,
What happens? . . . OUCH!
If the [5 Watt] piccolo is causing slew rate distortion of the amplifier, then your hearing probably is in danger.
Is that Music?
Perhaps some can consider the above.
They might see why an amplifier that has slew rate distortion of high frequencies at full power output, Can Sound Wonderful on Real Music.
Purchase an entry level digital oscilloscope, set it for FFT from low frequency to 25kHz, and set the trace to Infinite persistence.
Now, play one of your favorite recordings all the way through. Notice how the FFT shows amplitude dropping off as the frequency increases.
Still not sure, then play another favorite recording and look some more.
If the high frequencies are not dropping off at high frequencies, then perhaps . . .
Your amplifier is clipping
The recording is clipping
It is Electronic Music (perhaps Not Good electronic music).
You need to consider the complete process from the music performer all the way through equipment to the end (your ears).
It is rightly called a System.
Last edited:
I knew someone would find that.
But yeah, I moved on and now design and build far better amplifiers. I won't dissuade the OP but it seemed earlier in this thread he would build something interesting. Sadly this is not really the case.
No, just disappointed. But I guess it's all good. I started by designing more or less established circuits that worked and "sounded good" but with time I realized they didn't perform that great after all. I guess this is just the route you are taking. You can see from my old thread that Bugun found that I did this kinda thing too at one point. I thought I would save your some of the hassle. Guess not.
The 5842 should work. I prefer triode strapped pentodes because I got them when they were still affordable. Personally I would not use the monkey topology (again). I would call it a "compromise" circuit. But I will let you discover that for yourself! 😎
Or, you could simply do a classic design running 5842 at around 16mA with 160V on the plate using a 20k load resistor and a 125 ohm cathode resistor. Aim to bias around 2V on the cathode.
You will get better gain if you use a larger load resistor on your 5842. Don't think you will get an amplification factor of 55 with that little 4.5K load resistor.
Then use 3.5k ohm cathode resistor on the 2a3 - you will discover that you probably need something like 50uF cathode by-pass cap there. Your posted circuit is only running 2A3 at 50mA. You can push it up to 60mA and it will sound a lot better with lower distortion too.
You will find that 8-9mA is not going to be enough - unless you use a source follower with much lower output impedance. Sure it will sound better than the "wimpy" triodes, but still - go for the gusto! Test it out at least. 5842 is good for 4.5 Watts. Feel free to experiment around if that seems worth doing.
My current circuits only use Chokes and DC-link film caps in the power supply. Remember that the power supply is 100% in the pathway. Put a big noisy resistor in it? It's going to sound.
I am (still) using TV damper diodes for slow start-up too, but have also tried other slow start-up methods. You absolutely must have slow start-up for anything direct coupled.
Also, Mu-Follower is much less distorting than SRPP. you might want to consider that too.
Last edited:
One can still use the 6SF5 without any slew rate issue if a DC coupled cathode follower is added in the middle. Starting from the Komuro style circuit and using a proper input transformer it could be done without coupling capacitors. And it would be an integrated amplifier.
The input transformer has to be able to work with relatively high DC differential between primary and secondary. The new Hammond 1140 Studio Series can do that. I have asked info directly to their technical department. In this case one could use the 1:1 model 1140-LN-B and could put a 10K ALPS RK27 volume control across the secondary. Then have 2 supplies: +/- 270V DC (using a voltage multiplier to keep the power transformer simpler) and 250V + OPT primary drop for the 2A3. The 6SF5 has 1.5K bypassed cathode resistor and 320K anode resistor ( 1 mA anode current) and directly coupled to a half 6BL7GT cathode follower with 12K cathode resistor. Both these tubes are supplied with the dual +/-270V. This way the cathode of the 6BL7 should be around the right -43-44V bias for the 2A3 that will have its cathode grounded. The circuit is also safe on start-up as the grid of the 2A3 will see -270V until the cathode follower draws current. Of course, one can refine the supply as desired but without having to accept any worrying compromise. With different 2A3 operative point only small adjustments need to be done to both supplies. The dual supply could use a cap multiplier that also serves as fine adjustment for the actual output voltage or one could regulate both supplies with dual one still adjustable.
The cathode follower will bias at about 310V/19mA. It's near max for the tube but still ok (just under 12W for both plates) and should work in very linear conditions......
With lower anode voltage for the 6SF5 (it could be something like 180V instead of 210V and 1.2K cathode resistor) or using other lowish current (less than 5 mA) input tubes the circuit could be less "extreme" and possibly a 6SN7 cathode follower could be used.
N.B. The values for the various voltages and resistors are just a quick guess but should not be far off.....
The input transformer has to be able to work with relatively high DC differential between primary and secondary. The new Hammond 1140 Studio Series can do that. I have asked info directly to their technical department. In this case one could use the 1:1 model 1140-LN-B and could put a 10K ALPS RK27 volume control across the secondary. Then have 2 supplies: +/- 270V DC (using a voltage multiplier to keep the power transformer simpler) and 250V + OPT primary drop for the 2A3. The 6SF5 has 1.5K bypassed cathode resistor and 320K anode resistor ( 1 mA anode current) and directly coupled to a half 6BL7GT cathode follower with 12K cathode resistor. Both these tubes are supplied with the dual +/-270V. This way the cathode of the 6BL7 should be around the right -43-44V bias for the 2A3 that will have its cathode grounded. The circuit is also safe on start-up as the grid of the 2A3 will see -270V until the cathode follower draws current. Of course, one can refine the supply as desired but without having to accept any worrying compromise. With different 2A3 operative point only small adjustments need to be done to both supplies. The dual supply could use a cap multiplier that also serves as fine adjustment for the actual output voltage or one could regulate both supplies with dual one still adjustable.
The cathode follower will bias at about 310V/19mA. It's near max for the tube but still ok (just under 12W for both plates) and should work in very linear conditions......
With lower anode voltage for the 6SF5 (it could be something like 180V instead of 210V and 1.2K cathode resistor) or using other lowish current (less than 5 mA) input tubes the circuit could be less "extreme" and possibly a 6SN7 cathode follower could be used.
N.B. The values for the various voltages and resistors are just a quick guess but should not be far off.....
Last edited:
My post # 477 was not stated very well, so it did not make things very clear.
Hopefully, this Post will better explain an important principle:
Purchase a digital scope, and set it to FFT to cover 0Hz to 25kHz, and activate the scope's Max Hold function.
Connect the scope to the output of your CD player.
Put in a music CD that has lots of high frequency energy and play the CD all the way through.
You will notice that the higher frequencie's amplitudes drop off rapidly;
That is exactly according to natural fall-off rate of both the instrument's harmonics and the voice's harmonics.
Most of the energy is at low and medium frequencies, and those frequencies are at a much lower slew rate (more microseconds per sine wave).
This music is the exact signal that you are sending to your amplifier.
The higher the frequency, the more important slew rate is (Volts / microsecond). Agreed.
But, the higher the frequency, the lower the amplitude is of the signal is that comes from the CD player to your amplifier.
Again, Slew rate: "Volts / microsecond"
The term: / microsecond. The higher the frequency, the lower the microseconds is.
The term: Volts. The higher the frequency, the lower the Volts are from the CD player
The higher the frequency, the lower the slew rate is to your amplifier (the much lower Volts of the high frequencies)
We are talking about Music; we are not talking about a Full Amplitude 20kHz sine wave from the signal generator.
Concluding principle:
You Do need to be concerned about the slew rate of your power amplifier, But it does not have to slew 20kHz at full power output (Poor tweeter micro diameter voice coil).
Questions?
Hopefully, this Post will better explain an important principle:
Purchase a digital scope, and set it to FFT to cover 0Hz to 25kHz, and activate the scope's Max Hold function.
Connect the scope to the output of your CD player.
Put in a music CD that has lots of high frequency energy and play the CD all the way through.
You will notice that the higher frequencie's amplitudes drop off rapidly;
That is exactly according to natural fall-off rate of both the instrument's harmonics and the voice's harmonics.
Most of the energy is at low and medium frequencies, and those frequencies are at a much lower slew rate (more microseconds per sine wave).
This music is the exact signal that you are sending to your amplifier.
The higher the frequency, the more important slew rate is (Volts / microsecond). Agreed.
But, the higher the frequency, the lower the amplitude is of the signal is that comes from the CD player to your amplifier.
Again, Slew rate: "Volts / microsecond"
The term: / microsecond. The higher the frequency, the lower the microseconds is.
The term: Volts. The higher the frequency, the lower the Volts are from the CD player
The higher the frequency, the lower the slew rate is to your amplifier (the much lower Volts of the high frequencies)
We are talking about Music; we are not talking about a Full Amplitude 20kHz sine wave from the signal generator.
Concluding principle:
You Do need to be concerned about the slew rate of your power amplifier, But it does not have to slew 20kHz at full power output (Poor tweeter micro diameter voice coil).
Questions?
Last edited:
Yes, one can get other benefits that normally are not obtainable in other ways. Also in this case I am suggesting a 10K volume control which 5-10 times lower than usual. Using a ground lift at the primary, when there is no input the switch is closed and the primary is grounded; when connected to other equipment the primary is grounded into the source. This isolates current loops.
From Jensen site: " For eliminating hum and buzz in line-level signals, Faraday-shielded input transformers are the preferred solution. These transformers offer extended low-frequency and Bessel-tailored high-frequency responses, resulting in ultra-low time-domain distortion (DLP), and use internal Faraday shields for unmatched immunity to hum, buzz, and RF interference (high CMRR). Transformer-less inputs suffer “breakthrough” noise in electrically-hostile environments but transformers are unaffected. The controlled high-frequency response removes ultra-sonic “hard digital edge” artifacts from signals, resulting in a unmatched sonic clarity. Each transformer is enclosed in a MuMETAL® magnetic shielding can to reduce hum caused by any nearby magnetic fields. Several turns ratios are offered to optimize noise performance with various input stage designs. Extremely high dynamic range can be delivered, up to 140 dB, in real-world system environments."
The Hammond trafos have the same shielding and construction features regarding noise reduction.
I think I am going to give it a go soon but will use the 6C4C at 300V/60mA (about -56V bias) with 3.5K/8R for effective 5W output in A1 @ 4-5% THD.
+/-240V should be enough to step down to the 6SN7 as cathode follower and the 6SF5 working around 170-175V/1mA anode with 1.2K cathode resistor.
That's a verry important benefit of input transformer. There are other benefits, too. An input transformer makes perfect transition from balanced to unbalanced and unbalanced to balanced. It allows tweaking amplifier's gain structure. It provides low resistance path from input tube grid to ground, eliminating grid resistor nose. This is why input transformers were so common in pro equipment.
45,
Thanks for the idea!
Because of the 2-pole resonant characteristic of input transformers, plus the primary to secondary leakage reactance, then depending on the driving impedance and the loading impedance on the input transformer, they might have a Bessel rolloff function.
You make me think again of the Group Delay results.
A single pole low pass filter on the other hand, has a Gaussian rolloff function, with the classical Gaussian factor of 0.35.
Then -3 dB Bandwidth = 0.35/Rise Time. And the -1dB bandwidth is an octave lower in frequency.
All nice things to know, if you are looking for a dominant low pass filter.
Comment: This 2A3 amplifier thread is very long, that is OK.
But some people may be discouraged by some things that have been mentioned along the way; sometimes the comments criticize their design, and other factors such as parts availability or price get in the way of some builders.
To all of you that have, or want to have a good sounding 2A3 amplifier, I am reminded of a quote from a writer in Sound Practices Magazine:
. . . "I never met a 2A3 amplifier I did not like". Those are my experiences too.
Have fun Building and have fun Listening!
Thanks for the idea!
Because of the 2-pole resonant characteristic of input transformers, plus the primary to secondary leakage reactance, then depending on the driving impedance and the loading impedance on the input transformer, they might have a Bessel rolloff function.
You make me think again of the Group Delay results.
A single pole low pass filter on the other hand, has a Gaussian rolloff function, with the classical Gaussian factor of 0.35.
Then -3 dB Bandwidth = 0.35/Rise Time. And the -1dB bandwidth is an octave lower in frequency.
All nice things to know, if you are looking for a dominant low pass filter.
Comment: This 2A3 amplifier thread is very long, that is OK.
But some people may be discouraged by some things that have been mentioned along the way; sometimes the comments criticize their design, and other factors such as parts availability or price get in the way of some builders.
To all of you that have, or want to have a good sounding 2A3 amplifier, I am reminded of a quote from a writer in Sound Practices Magazine:
. . . "I never met a 2A3 amplifier I did not like". Those are my experiences too.
Have fun Building and have fun Listening!
Last edited:
For those that don't have a scope with spectrum analysis, you can see this effect without any hardware, other than means to loop your PC soundcard (line out → line in).
Take your digital audio music file, and play it (With Audacity for better visualisation of the time domain of the music).
You can then use Arta software to perform the FFT spectrum analysis.
The Max. hold function is achieved by Arta's Avg → Peak Hold. Increase the number of Peak Hold samples to cover more of the Track, if desired.
The screenshot is a hundred 16k FFT samples (@44100 smpl/s) of the 'Invasion Theme' (D. Shostakovich: Symphony №7: I. )
Arta can be installed in Linux, with WINE, for a somewhat-FOSS solution. (launch Arta with sudo, for USB soundcards).
Last edited:
Personally, I use a USB soundcard with some simple protection circuit added. I wouldn't want to fry the soundcard on my laptop - its on the motherboard.
I can vouch for using ARTA. The price is quite reasonable considering what you get. Thanks for this post Rod!
I can vouch for using ARTA. The price is quite reasonable considering what you get. Thanks for this post Rod!
It's OK.
this test doesn't measure the tube amplifier. It justs connects the PC's (analogue) line out to line in. I just put a 3.5mm male→male jack lead across my old desktop's audio card.
The playback of the audio file simply routes to the FFT software, which is waiting for it at line in. It's safe to use with internal or external cards, though I suspect you may need a USB soundcard to get analogue signals in today's laptops.
Please forgive my attempt at a rough sketch on drawing 45's circuit. I added a bias pot to dial in the bias voltage. It's rough but the concept is in there. I forgot to draw the filament transformer for the 2A3 but you can add your favorite AC or DC supply.
Please share "The Trick" about circuit regarding the batteries. Today I use battery bias by placing the battery (+) to the filament pin and the battery (-) to ground. Simple cathode bias. Do you advise a different method?