I'm wondering, what would be the effective model of an amplifier when it's in the process of "slewing", vs. when it's at it's Q-point? If the amplifier has significantly less open loop gain in this state, it could certainly mean reduced fidelity. Just as distortion worsens at high frequencies, when many amps are beyond their dominant pole and gain and hence the amp's ability to control distortion, are reduced.
If this has been covered in an article someone knows about I'd be happy to give it a read.
If this has been covered in an article someone knows about I'd be happy to give it a read.
If one goes to the trouble of simulating the full environment that an amplifier has to handle, like poor quality mains being fed to a far less than perfect power supply, with the componentsbeing joined together with parasitic behaviours in the links between them, and back to the power supply then one marvels that amplifiers actually sound as good as they do - it's easy to see the circuit struggling, trying to deal with all the distortion inducing "imperfections".
A simple example: a relatively straightforward design, plenty of global feedback, pretty good distortion figures when using ideal power supplies - replace the latter with something far closer to real voltage rails ... and the distortion immediately is something like 100 times worse. Why? Because the feedback has to handle both the circuit not being linear AND the power supply modulating at the same time.
Theory may be great as an intellectual exercise, but at some stage reality hits back ...
A simple example: a relatively straightforward design, plenty of global feedback, pretty good distortion figures when using ideal power supplies - replace the latter with something far closer to real voltage rails ... and the distortion immediately is something like 100 times worse. Why? Because the feedback has to handle both the circuit not being linear AND the power supply modulating at the same time.
Theory may be great as an intellectual exercise, but at some stage reality hits back ...
Jan Lohstroh who designed the "Otala Amplifier" is rather impartial to the acclaim his creation has received from elements within the audiophile community.
http://www.linearaudio.nl/linearaudio.nl/images/pdf/Jan_Lohstroh_interview(1).pdf
Hell, Frank, all you need to do is to increase the value of the Miller capacitor, you know, the one I don't know anything about, to a high value, and in the old designs, you could easily see values like 220 pF, even 330pF, if memory serves, and give it relatively fast input signal, yet low enough not to cause saturation or clipping. It's that simple. Mind your ears though.
If you "objective" scientist wanna-bes could be just 10% as objective as you claim to be but never are, and think outside the routine for 10 seconds, you might cocnclude that the Otala/Lohstroh amp and the theories behind it in fact DID change the audio world. No they were not right all the way, they exaggerated some things and possibly blew them out of proportion, but they did change things. As the prime result of that work, designers started paying much more attention to TRANSIENT distortion effects, which did produce some rather good amps of the time.
If nothing else, they made people sit back and think. This is, as far as I can tell, immesurably more than any of the naysaysers ever did. I never claimed oTala was the neginning and the end, instead I read his other papers on driving difficult loads and I followed his own evolution during his Harman/Kardon years. Trust me, it's not hard, it's all there, you just have to look a bit.
As it is, far too many people devote too much of their time trying to prove that Otala was wrong, rather than trying to contribute to the pool of knowledge with something new. And I find most naysayers to be as radical as the Islamic State fedayeens in trying to prove Otala wrong.
You colse your eyes to the fact that modern design practices have more of Otala in them than you'd like. For example, while he was not the first to do so (e.g. Bernard Kardon had been at it 15 years before the Otala amp, and I'm sure he wasn't the only one), he WAS the one who established wide bandwidth as a standard design practice. For confirmation, look at the specs of amps before and after him. Today, nobody thinks of it much, it's there, period.
I'd like to see some of your work which might demonstrate just how wrong he was and where, referred to his master work, the Harman/Kardon Citation XX. Let's see you push 200 A into a loudspeaker, even in peaks.
Feedback a cure for FET Ron? Not really. Multiple devices at sensible bias works. I had a thought on that and it could be a very good one. If FET's no 1 & 2 were at > 100 mA bias and the rest zero bias we might have something. Rob Elliot shows how zero bias FET's have good distortion up to 10 kHz. Bias extends this to 100 kHz. One thing that works very well with feedback is to make the moment to moment changes very small. The linearity is of zero importance really. If you take the bipolar transistor it is a switch as in a light switch. The FET more like a dimmer. The beauty of the FET is it offers something for the feedback to work with at any level. If biased high enough the nasties come it at already loud levels. As the low bias and high bias curves are curves the composite curve should be OK. How hard is it to set up? As simple as electronics age 12 at school. No bias drift and no real SOA problems. 1 & 2 driven by fixed resitors with a centre split for the zero bias ones. Unlike bipolar zero bias FET is nothing of the sort. It is minimal bias, there still is a curve. This would be Audio FET's like Exicon or BU900 etc.
My baffles are now dry. They have a 7 mm curve of equal shape on both ( sits in the other exactly ). Looks deliberate. I think I will keep it. The MDF side convex. As the ply looks nice the possibly worse concave surface will be used. If 2 x 3 ply was used with MDF filling that might be the best. If so glue in one hit. The 3 buckets of water works a treat as weights. The last 4 inches didn't glue. I got the glue in and used 1 inch clamps ( £ shop ). Two engineers 2 foot rulers to avoid marks. 7 clamps used. Any thoughts on beaming? I suspect the 12 Lta driver will be the major cause of polar problems and not a gentle curve?
The reason I chose the tripples as I like to call them is the complexity of the triple tripple if you like is not much more than the tripple MOS FET. One Vbe transistor, two pre driver and 2 drivers. The cost of the devices in favour of the all NPN version. Or all PNP if really wanting to save money. PNP devices are for the moment cheaper. They are reducing stock I guess. All PNP was the typical germanium option and the Goodmans Module 80 was all PNP silicon ( 2955 ). It is supected it was a gemanium design made larger in a hurry. I don't completely see the point of a tripple triple if NPN and PNP. I would prefer an over sized VAS and Darlingtons or better still complimentary feedback pairs. If care is taken the driver alone can drive triple devices. As the realistic output is 10 amps sine wave for all 3 devices that says the drive current should be about < 1 amp. If wanting transient power the drive should be about 3 amps peak as a good guess. This also speeds up the stage. One almost wins the gain lost by using multiple devices. The 2N3055 will drive 1 ohms and give 50 watts. It has if lucky a gain of 5 if so. It needs a 3055 to drive it. At 3 amps is will have a gain more like 50. Thus 3 will have higher gain. The optimum resistive load for a transistor was thought to be 5R6. If my rule of thumb is workable that is 10 R reactive. 3R3 for 3 devices.
My baffles are now dry. They have a 7 mm curve of equal shape on both ( sits in the other exactly ). Looks deliberate. I think I will keep it. The MDF side convex. As the ply looks nice the possibly worse concave surface will be used. If 2 x 3 ply was used with MDF filling that might be the best. If so glue in one hit. The 3 buckets of water works a treat as weights. The last 4 inches didn't glue. I got the glue in and used 1 inch clamps ( £ shop ). Two engineers 2 foot rulers to avoid marks. 7 clamps used. Any thoughts on beaming? I suspect the 12 Lta driver will be the major cause of polar problems and not a gentle curve?
The reason I chose the tripples as I like to call them is the complexity of the triple tripple if you like is not much more than the tripple MOS FET. One Vbe transistor, two pre driver and 2 drivers. The cost of the devices in favour of the all NPN version. Or all PNP if really wanting to save money. PNP devices are for the moment cheaper. They are reducing stock I guess. All PNP was the typical germanium option and the Goodmans Module 80 was all PNP silicon ( 2955 ). It is supected it was a gemanium design made larger in a hurry. I don't completely see the point of a tripple triple if NPN and PNP. I would prefer an over sized VAS and Darlingtons or better still complimentary feedback pairs. If care is taken the driver alone can drive triple devices. As the realistic output is 10 amps sine wave for all 3 devices that says the drive current should be about < 1 amp. If wanting transient power the drive should be about 3 amps peak as a good guess. This also speeds up the stage. One almost wins the gain lost by using multiple devices. The 2N3055 will drive 1 ohms and give 50 watts. It has if lucky a gain of 5 if so. It needs a 3055 to drive it. At 3 amps is will have a gain more like 50. Thus 3 will have higher gain. The optimum resistive load for a transistor was thought to be 5R6. If my rule of thumb is workable that is 10 R reactive. 3R3 for 3 devices.
OB SPEAKERS
The MDF+plywood hybrids working well with the 12 Lta. The low mass route was better, I will try to make it work. Problem is it brakes easilly and has no strength for screws to hold fast to. The driver now sounds like a good PA unit which what it is. Voice quality is still stuning. Sound of reality outside the room as before. Just more shouty and shows the frequency range is restricted more than it did. Really great on TV. Loves EQ and even seems to have LF and HF if so. Better than Bose 901 I would say and far more open. Even though Eq'ed like crazy it is half the volume of the SMG'a. That suggests the 99 dB/ watt is an underestimate. Voice might just be better than the SMGa. If so I suspect the balance is more like studio monitors. There is a slight chestiness which I suppose is the ply. The Spendor BC1 I repaired were very similar on that. The Spendor had no greater bass and booomed more. On balance the BC1 is better. Only because it makes fewer mistakes. The sound is 1960's cinema with 2015 sources to compliment it. That works very well. LP12 is OK with them. Mostly this is a high-end sound for very little money. The drivers ignore the EQ, they thrive on it. The Tannoy is said not to.
One thing to make clear. The EQ is acive. There is no loss of amplifer coupling due to inductors or crossovers. I have extended the amplifer coupling to 4.2 Hz ( 16 Hz when the SMGa when standard ). This is running in the output caps for the Quad clone.
I must have spent 100's of hours thinking about this over 25 years ( Since Robin Marshall explained baffles to me ). The doing it could be 3 hours maximum. The cost about £120 ( $180 ) total is from Amazon for the drivers and Wicks builders merchants. These speakers prove to me OB is best and what we hear wrong in speakers is box then wood. I used £4 of glue. Use your hands to spead the glue if PVA. So quick and makes it unlikely it will be too thin. I spread the glue in 5 minutes then tapped it to get best look, the glue makes it slide nicely. The water bucket weights the largest ones.
When I have steamed the ding marks out and Danish oiled them I will post a photo. The builders merchant put the dings in. I chose planks with a nice grain and knew perfection was not part of the mix.
My next design chalenge is the feet. It will be what the Loricraft scrap heap has that will be the choice. Some mild steel would be OK as the SMGa have.
The MDF+plywood hybrids working well with the 12 Lta. The low mass route was better, I will try to make it work. Problem is it brakes easilly and has no strength for screws to hold fast to. The driver now sounds like a good PA unit which what it is. Voice quality is still stuning. Sound of reality outside the room as before. Just more shouty and shows the frequency range is restricted more than it did. Really great on TV. Loves EQ and even seems to have LF and HF if so. Better than Bose 901 I would say and far more open. Even though Eq'ed like crazy it is half the volume of the SMG'a. That suggests the 99 dB/ watt is an underestimate. Voice might just be better than the SMGa. If so I suspect the balance is more like studio monitors. There is a slight chestiness which I suppose is the ply. The Spendor BC1 I repaired were very similar on that. The Spendor had no greater bass and booomed more. On balance the BC1 is better. Only because it makes fewer mistakes. The sound is 1960's cinema with 2015 sources to compliment it. That works very well. LP12 is OK with them. Mostly this is a high-end sound for very little money. The drivers ignore the EQ, they thrive on it. The Tannoy is said not to.
One thing to make clear. The EQ is acive. There is no loss of amplifer coupling due to inductors or crossovers. I have extended the amplifer coupling to 4.2 Hz ( 16 Hz when the SMGa when standard ). This is running in the output caps for the Quad clone.
I must have spent 100's of hours thinking about this over 25 years ( Since Robin Marshall explained baffles to me ). The doing it could be 3 hours maximum. The cost about £120 ( $180 ) total is from Amazon for the drivers and Wicks builders merchants. These speakers prove to me OB is best and what we hear wrong in speakers is box then wood. I used £4 of glue. Use your hands to spead the glue if PVA. So quick and makes it unlikely it will be too thin. I spread the glue in 5 minutes then tapped it to get best look, the glue makes it slide nicely. The water bucket weights the largest ones.
When I have steamed the ding marks out and Danish oiled them I will post a photo. The builders merchant put the dings in. I chose planks with a nice grain and knew perfection was not part of the mix.
My next design chalenge is the feet. It will be what the Loricraft scrap heap has that will be the choice. Some mild steel would be OK as the SMGa have.
Try engine valve springs...seriously.
Add a web of RTV silicone rubber between coils to add a bit of damping.
Dan.
Last edited:
Max. I went to a RAF college so spot on.
Tannoy made a design for a wartime contract under the understanding that the name Tannoy would always be seen. It has become a generic word for PA system.
My dad was RAF and was never allowed to leave. I suspect technically although 84 he still can be called up at a moments notice. He had very specail skills which I am not allowed to talk about. He was in electronics.
Tannoy made a design for a wartime contract under the understanding that the name Tannoy would always be seen. It has become a generic word for PA system.
My dad was RAF and was never allowed to leave. I suspect technically although 84 he still can be called up at a moments notice. He had very specail skills which I am not allowed to talk about. He was in electronics.
What resonant frequency and Q do you think is best for the plinth/spring system?
Like the bad boy runner from South Africa I guess? That was the idea so thanks for the nudge. I will clone the SMGa I think for now. Our blacksmith Nimbo might have some. Problem with Nimbo is never do today what can be done never. If I can get the metal out of him that will be OK. 1/4 inch looks right. The grinder needs new bearing and makes stuff jump. Health and safety, what what hey hey ?
In past experiments Ford V8 valve springs supporting decently heavy floor standing speakers gave resonance at around 0.5 hz or so.
Ditto using offcuts of heavy slab of kitchen top marble ...one sheet of marble under cdp, and one on top of the cdp, all supported on four valve springs.
The effect was to 'float' the cdp/speakers, providing really good isolation from the wooden house/wooden floor.
Depending on the mass of your gear, Mack truck or Hyundai springs might be best....see your local mechanic/wrecking yard.
Very low system resonance frequency (well below audio band frequencies) I think is optimal, and adding the silicone webbing shuts up spring self resonance.
Dan.
Last edited:
I'll have to give it a try. Even though my basement is right on slab, you might have seen the picture on the food thread as to just how close the commuter rail is. Several of my LP transfer sessions have been ruined by an inbound train. The railroad ran out of money and could only convert one direction to continuous rail.
There was a problem in the earlier days of transistor amplifiers in getting too greedy in terms of the distortion figures at the expense of amplifier stability.
In an overload situation where the negative feedback loop is interrupted the recovery effects will be discernable.
Apart from amplifier clipping, this situation could arise when a voltage amplifier stage could not track the input signal.
There were ways to mitigate the risks but distortion percentages were a prominent measure in advertisements.
On the other hand stability margins received no mention in sales brochures. Had this been otherwise it would have required some technical knowledge on the part of shop sales staff whereas confusing the customer could hinder making a sale.
Since loudspeakers comprise a reactive load these can induce a signal into the feedback loop and it is unsafe to set stability allowances assuming a restive load.
If an amplifier is operated with insufficient standing current through the output stage there will be a permanent region of operation at the crossover point where there is zero feedback.
These are all bogeys of the past.
Oh boy, and I thought I was bad off with 4 bus and 3 tram lines just underneath my apartment building (I'm on the 8th floor) and now I see I'm not really bad off.
We live in a world economy with consumer watchdogs and guarantee requirements. There are accepted standards for Hi-Fi as you mentioned in a recent post and magazines whose purpose is to inform the public on matters of choice. It is hard to believe there are manufacturers who would risk their reputations and loss of sales in a very competitive market by making an unsatisfactory product.
Anything is possible on the DIY scene so sharing of information is important.
Fortunate enough to never have to worry about the earth moving for me, - but a key part of what I do is to always make sure that the vibrational energy of the speaker carcase can be drained off, and largely nulled by being coupled to high mass. An absolutely key part of my original good sound setup was quite substantial mass loading of the very ordinary, lightweight speaker cabinets - this gave the very straightforward mid/woofer drivers the ability to produce highly authoritative, deeply felt bass lines - I hear big mutha subwoofers now and again these days, and often find them quite laughable - as fake as "red faces and nuclear grass" on a TV set ...
- but a key part of what I do is to always make sure that the vibrational energy of the speaker carcase can be drained off, and largely nulled by being coupled to high mass. An absolutely key part of my original good sound setup was quite substantial mass loading of the very ordinary, lightweight speaker cabinets - this gave the very straightforward mid/woofer drivers the ability to produce highly authoritative, deeply felt bass lines - I hear big mutha subwoofers now and again these days, and often find them quite laughable - as fake as "red faces and nuclear grass" on a TV set ...
Last edited:
- Status
- Not open for further replies.