I'm good Jeff, looks like I had a DVM issue, used another and I'm at 1.56 on both sides.
I am getting some rail sag though... I assume you always get some, but under load they are about 21v on both sides. Is that normal, or am I running out of Trans? 300VA 18V Antek
I'm actually listening to some Nora Jones through the Yarra MlB to the J2 to Takton DIs... sounds pretty good on 4R speakers :O. The volume on the Yarra is barely cracked and it's running 80db 3 foot from the speakers. Yikes!
I am getting some rail sag though... I assume you always get some, but under load they are about 21v on both sides. Is that normal, or am I running out of Trans? 300VA 18V Antek
I'm actually listening to some Nora Jones through the Yarra MlB to the J2 to Takton DIs... sounds pretty good on 4R speakers :O. The volume on the Yarra is barely cracked and it's running 80db 3 foot from the speakers. Yikes!
Last edited:
Great news on the bias!
The chart about 1/3 down this page: Transformers Part 2 - Beginners' Guide to Electronics shows that a 300VA transformer typically has a regulation around 7%. 18VAC = 25.4VDC - 1.3V rectifier drop = 24.1V. So I'd expect to see a no-load voltage after the rectifiers of 25.8V, dropping to 24.1V under load.
Then there are the resistors in the CRC filter. If you're drawing about 3A and using 4 0R47 resistors then that's another 0.4V lost, bringing your rails to ~ 23.7 under load. So something's not quite right....
Take an AC measurement before the rectifiers and a DC measurement between the rectifiers and the CRC filter and a DC measurement on the amp board and report back.
The chart about 1/3 down this page: Transformers Part 2 - Beginners' Guide to Electronics shows that a 300VA transformer typically has a regulation around 7%. 18VAC = 25.4VDC - 1.3V rectifier drop = 24.1V. So I'd expect to see a no-load voltage after the rectifiers of 25.8V, dropping to 24.1V under load.
Then there are the resistors in the CRC filter. If you're drawing about 3A and using 4 0R47 resistors then that's another 0.4V lost, bringing your rails to ~ 23.7 under load. So something's not quite right....
Take an AC measurement before the rectifiers and a DC measurement between the rectifiers and the CRC filter and a DC measurement on the amp board and report back.
Mains is 14 leaving the trans is just the wire that Antek uses directly to the rectifiers. Out of the recs is 16.
All sounds good -- although I couldn't tell you where the missing volt-and-a-half are.
It's probably time for "don't worry; be happy".
Ya, but I want that volt, or in this case 2... 20v trans on the way.
Thanks again for your extreme patience, I will read the link when I get time. Still a couple books too. "I have miles to go before I sleep."
I am enjoying it though... Dire Straights at the moment.
Cheers Jeff
JT
The conventional rectifiers that I have used have 'eaten' between 2.8V and 3.5V from the 'ideal' voltage that would be predicted in a perfect world. The more common GBPC3502 blocks tend toward 3.5V, and it takes a set of FEP30 diodes to get down around 2.8V. The Vishay VS-26MB40A block rectifiers and LVB2650 inline bridge rectifiers fall in between these. This is measuring from the AC inputs of the bridge to the + and – outputs. I haven't seen more than 0.2V sag from the secondary windings, as long as the transformers are properly sized. For our purposes of building FW clones, that means a single 400VA transformer, or better yet a pair of 300VA transformers.
So the world is not perfect, what else is new? Expecting to use the simple formulae for voltage drop from a bridge rectifier will likely result in disappointment. Even the newer synchronous rectifiers don't yield the 'Ideal' results, but they are an improvement over conventional diodes.
So the world is not perfect, what else is new? Expecting to use the simple formulae for voltage drop from a bridge rectifier will likely result in disappointment. Even the newer synchronous rectifiers don't yield the 'Ideal' results, but they are an improvement over conventional diodes.
Thanks TA,
The last three, only three I built, all were as expected, this seems to be the odd ball. Anyway, it's good to hear that this in not uncommon, and indeed something that can be expected... I guess, it easy to get a T with more capacity, because you can always reduce, but not gain.
JT
The last three, only three I built, all were as expected, this seems to be the odd ball. Anyway, it's good to hear that this in not uncommon, and indeed something that can be expected... I guess, it easy to get a T with more capacity, because you can always reduce, but not gain.
JT
There are usually several factors which contribute to the rectified voltage in a conventional linear power supply. This article remains one of the best that I have found to explain how rectification works: The Valve Wizard The first few paragraphs are general, while much of the later material is more specific to high-voltage tube power supplies.
One of the first things to remember is the definition of RMS voltage when measuring AC voltage performance of transformers, and how multimeters operate. If we were to measure the AC power voltage at the primary inputs of a power transformer (be careful!), a reading of 120V means an RMS amplitude of 120. Similarly, the voltage on the secondary windings is also specified and measured as an RMS amplitude. This is definitely not the same as peak-to-peak voltage, which is actually twice the (non-RMS) amplitude. That is how we get the formula: Vds = 1.4 x Vrms
But we know that is an oversimplification. Even a factor of 1.3 doesn't really work very well, especially at the lower voltages used in solid-state amplifiers. (The higher voltages used for tube power supplies, and the variations in voltage sag of the relatively under powered transformers tend to swamp out the differences between one or two diode drops.)
In addition to that, the parameters that are used to characterize rectifiers could also use come clarification. The forward voltage (Vf) figure of merit does, in fact, refer to a single diode in the bridge. However, it only represents a data point taken under certain conditions, as diodes are exponential curve devices, at best, and are highly non-linear. That means that the diode drop function of a sum of currents is not equal to the sum of the individual functions. Or, F(a+b) <> F(a) + F(b).
So when we are attempting to apply that approximate formula for Vdc, adding two diode drops to account for the bridge rectifier still doesn't give an accurate prediction. The engineering solution, of course, is to add a fudge factor to account for the actual bridge rectifier drop. And that fudge factor can only be determined empirically by taking actual measurements, rather than relying on data sheets. Note that the fudge factor will depend on a sufficient amount of bulk capacitance on the output of the rectifier.
One of the first things to remember is the definition of RMS voltage when measuring AC voltage performance of transformers, and how multimeters operate. If we were to measure the AC power voltage at the primary inputs of a power transformer (be careful!), a reading of 120V means an RMS amplitude of 120. Similarly, the voltage on the secondary windings is also specified and measured as an RMS amplitude. This is definitely not the same as peak-to-peak voltage, which is actually twice the (non-RMS) amplitude. That is how we get the formula: Vds = 1.4 x Vrms
But we know that is an oversimplification. Even a factor of 1.3 doesn't really work very well, especially at the lower voltages used in solid-state amplifiers. (The higher voltages used for tube power supplies, and the variations in voltage sag of the relatively under powered transformers tend to swamp out the differences between one or two diode drops.)
In addition to that, the parameters that are used to characterize rectifiers could also use come clarification. The forward voltage (Vf) figure of merit does, in fact, refer to a single diode in the bridge. However, it only represents a data point taken under certain conditions, as diodes are exponential curve devices, at best, and are highly non-linear. That means that the diode drop function of a sum of currents is not equal to the sum of the individual functions. Or, F(a+b) <> F(a) + F(b).
So when we are attempting to apply that approximate formula for Vdc, adding two diode drops to account for the bridge rectifier still doesn't give an accurate prediction. The engineering solution, of course, is to add a fudge factor to account for the actual bridge rectifier drop. And that fudge factor can only be determined empirically by taking actual measurements, rather than relying on data sheets. Note that the fudge factor will depend on a sufficient amount of bulk capacitance on the output of the rectifier.
Last edited:
Jeff, I something, I think, is fairly important to share. Remember you commented on how close the one board was to the sink? That allows for a better placement of the JFET on the sink... ergo, not partially hanging off, or at the least up closer to the main sink.
The reason why I bring this up is I mounted the second per your template/PCB instead of drilling new holes. It's much neater to use the template or the PCB and I commented that I wish I had done this with the first one.
Okay, so I'm been looking at temps with the odd mounted facing the bench and the amp on its side and no cover. There is a fan moving some air in the room and the top side with fins in the breeze still runs almost 10c hotter than the lower side getting no air to the fins. At first I was puzzled, but then as I looked at it I understood. By moving the FETs in by just less than an 3mm, makes the difference. The two FETs I have almost touching the heat-sink are running at 45C while the ones that are mounted further out are running 64-66C
Just an FYI for you and anyone doing the build... I think if you didn't care about having the, "J2 look," and mounted the FETs flat on a sink, you might get away with a lot less sink, or at least some much lower temps on the FETs.
The reason why I bring this up is I mounted the second per your template/PCB instead of drilling new holes. It's much neater to use the template or the PCB and I commented that I wish I had done this with the first one.
Okay, so I'm been looking at temps with the odd mounted facing the bench and the amp on its side and no cover. There is a fan moving some air in the room and the top side with fins in the breeze still runs almost 10c hotter than the lower side getting no air to the fins. At first I was puzzled, but then as I looked at it I understood. By moving the FETs in by just less than an 3mm, makes the difference. The two FETs I have almost touching the heat-sink are running at 45C while the ones that are mounted further out are running 64-66C
Just an FYI for you and anyone doing the build... I think if you didn't care about having the, "J2 look," and mounted the FETs flat on a sink, you might get away with a lot less sink, or at least some much lower temps on the FETs.
Last edited:
Interesting. Can you measure the temps of the heatsinks right next to the JFETs and report the difference? I'm curious if it's greater case-to-sink resistance or inner-sink-to-outer-sink resistance.
Cheers,
Jeff.
Ahhh... indeed I was missing the second half of that. (Which I guess in hind-sight should have been obvious as Vbe also varies over temp, current, etc.)
Thanks for the info (and the link).
Cheers,
Jeff.
Okay, looks like I need to recant. In making these measurement for you I notice an anomaly when the thermo was turned on it's side. WHen shooting it perpendicular the reading are much closer... maybe 3-4c dif in the two rails, the closer still being better, but not to the extreme I measured before. It looks like there an issue with this gun when turned in a parallel orientation with the rail. Since the bottom rail is on the bench I naturally rotated the gun when taking those readings. It was a fluke that I happened to turn the gun perpendicular and then upside down which then showed the numbers much closer.
The sink temp is slightly hotter on the closer one, but after looking at the data, I would say you might get 2-3C better with them mounted closer, but that might be within the margin of error with the test tool.
I'm glad you had me check the sink rails, in doing so I changed the orientation of the tool and was shocked to see a 10C change in the readings.
Last edited: