What is the importance of power/insertion/copper loss in output transformers?
I see that some really great iron is flat out to 60-100khz. However I keep seeing this insertion loss spec. What is the importance of this spec and what is the effect on sound quality? Does it just mean that I will get a few less watts output?
I also assume that when I read copper loss, insertion loss, and power loss they all mean the same thing.
I see that some really great iron is flat out to 60-100khz. However I keep seeing this insertion loss spec. What is the importance of this spec and what is the effect on sound quality? Does it just mean that I will get a few less watts output?
I also assume that when I read copper loss, insertion loss, and power loss they all mean the same thing.
afaik insertion loss is a generic term applicable to every component of any signalling system.
Connectors have an insertion loss spec. Splices have an insertion loss spec.
In 1999 i was working for an experimental broadband ISP and they had one of our fiber techs show us how he does his thing. Whole toolbag of equipment, careful process, sandpaper for polishing the end of the fiber that looks and feels glossy.
He proudly showed us on his $10,000 fiber analyzer that he was achieving iirc 3db insertion loss on a splice vs. the industry standard 5db, because he's good at his job.
2007 i had fiber run to my new house for gigabit internet access (though I'm provisioned at 150mbps). A much younger tech showed me the $5000 device that does the whole splice job for him. Insert fibers and press button. And explained that industry standard is now 1.5db, but this box averages 1.3db.
Your connectors have an insertion loss rating (if they are any good). A length of wire has an insertion loss coming from a combination of ohmic resistance, inductance, and capacitance (yeah, even a wire has capacitance with it's own insulation and it's neighbors). A capacitor has an insertion loss. etc. it's just a way of saying "This much signal doesn't come out the other side".
I wouldn't say that insertion loss is the same thing as copper loss - I'd say that copper loss is one form of insertion loss. In a transformer with a metallic core I'd assume that some of the insertion loss is from the windings and some of it is from energy wasted as heat or vibration in the core, so i don't think i fully understand the term "copper loss" as amplifier builders understand it.
Power loss would be a function of insertion loss.
Connectors have an insertion loss spec. Splices have an insertion loss spec.
In 1999 i was working for an experimental broadband ISP and they had one of our fiber techs show us how he does his thing. Whole toolbag of equipment, careful process, sandpaper for polishing the end of the fiber that looks and feels glossy.
He proudly showed us on his $10,000 fiber analyzer that he was achieving iirc 3db insertion loss on a splice vs. the industry standard 5db, because he's good at his job.
2007 i had fiber run to my new house for gigabit internet access (though I'm provisioned at 150mbps). A much younger tech showed me the $5000 device that does the whole splice job for him. Insert fibers and press button. And explained that industry standard is now 1.5db, but this box averages 1.3db.
Your connectors have an insertion loss rating (if they are any good). A length of wire has an insertion loss coming from a combination of ohmic resistance, inductance, and capacitance (yeah, even a wire has capacitance with it's own insulation and it's neighbors). A capacitor has an insertion loss. etc. it's just a way of saying "This much signal doesn't come out the other side".
I wouldn't say that insertion loss is the same thing as copper loss - I'd say that copper loss is one form of insertion loss. In a transformer with a metallic core I'd assume that some of the insertion loss is from the windings and some of it is from energy wasted as heat or vibration in the core, so i don't think i fully understand the term "copper loss" as amplifier builders understand it.
Power loss would be a function of insertion loss.
In a good quality output transformer the insertion loss is mainly the copper loss.
Losses in the core can also contribute to insertion loss but as long as the core excitation is low enough (under 1 Tesla in a good transformer) the core loss is not an issue.
The lower the DC resistances of primary and secondary windings, the lower the insertion loss will be.
Highish DC resistances will not only cost some output power but in my experience the main effect of a low DC resistance output transformer is a more dynamic sound.
Losses in the core can also contribute to insertion loss but as long as the core excitation is low enough (under 1 Tesla in a good transformer) the core loss is not an issue.
The lower the DC resistances of primary and secondary windings, the lower the insertion loss will be.
Highish DC resistances will not only cost some output power but in my experience the main effect of a low DC resistance output transformer is a more dynamic sound.
So my understanding from your replies is that it does not mean there is less power. A higher insertion loss means a loss of micro-details? Pieter t, I am lacking an electrical background as I work work in IT and this is more of a hobby. Not sure if I fully understood the impact on sound or how to tell if the core loss is acceptable.
I can see that Magnequest has an average of 0.5-0.6 db insertion loss on average while their best S-271-A has 0.3db insertion loss.
Japanese 100+ watt OPT such as Hashimoto, Tango, and Tamura avg about 0.2-0.3 db loss and is incredibly expensive. Hopefully their prices are cheaper in Japan as my family is planning a trip.
http://www.asahi-net.or.jp/~UP2J-KNST/tube/tb-trans1.htm
I am quite attracted to Lundahl for their prices at K&K audio however many of their power losses are higher when a low primary impedance is needed. For example their LL1623 will give 125W @ 3.0K 150H with power losses of 0.5db. And their LL1620 will give 180W @ 3.3K 300H but with power losses as high as 0.8db. Lundahl's power losses are much lower in all output transformers at 4.5K+ but I need a low primary impedance for my project (3.3K is ideal).
I can see that Magnequest has an average of 0.5-0.6 db insertion loss on average while their best S-271-A has 0.3db insertion loss.
Japanese 100+ watt OPT such as Hashimoto, Tango, and Tamura avg about 0.2-0.3 db loss and is incredibly expensive. Hopefully their prices are cheaper in Japan as my family is planning a trip.
http://www.asahi-net.or.jp/~UP2J-KNST/tube/tb-trans1.htm
I am quite attracted to Lundahl for their prices at K&K audio however many of their power losses are higher when a low primary impedance is needed. For example their LL1623 will give 125W @ 3.0K 150H with power losses of 0.5db. And their LL1620 will give 180W @ 3.3K 300H but with power losses as high as 0.8db. Lundahl's power losses are much lower in all output transformers at 4.5K+ but I need a low primary impedance for my project (3.3K is ideal).
Last edited:
You get less power!
Insertion loss in output transformers is (implicitly) referred to 1KHz. So it's just DC copper loss. Otherwise one should consider core losses and AC resistance as well. Cheap transformers are generally bad in terms of AC resistance (i.e. the resistance of the windings increases quite a lot with frequency and so losses).
I agree with Pieter that insertion loss has an impact on the sound. Transformers with insertion loss below 0.3 dB are generally said to be "transparent".
P.S.
C-cores transformers generally allow better figures for copper losses.
Lundahl's might also be getting an additional small advantage because not using a coil former (so might have a bit more space and thus use slightly larger wire). However this is only true for the higher primary impedance. I always avoid 0.8 dB connection with these transformers because in many cases frequency response is poor, especially for SE application where they don't even do 20KHz....
C-cores transformers generally allow better figures for copper losses.
Lundahl's might also be getting an additional small advantage because not using a coil former (so might have a bit more space and thus use slightly larger wire). However this is only true for the higher primary impedance. I always avoid 0.8 dB connection with these transformers because in many cases frequency response is poor, especially for SE application where they don't even do 20KHz....
No. Loss means less power.Pegasus123 said:
Unlikely. Loss means you get out less then you put in; it doesn't necessarily mean you get out something different from what you put in.
There are two main sources of loss in a transformer: copper resistance in the windings, and core losses due to hysteresis and eddy currents. Power transformers are usually designed so these losses are roughly equal. Audio transformers may be different.
Power loss in output transformers is referred to 1KHz and resulting from DC resistance (i.e. AC resistance is neglected because usually small in proper designs).
But it is quite likely that you get out something different from two transformers with same power loss. They can easily have different AC resistance. AC resistance is frequency dependent and already from 5KHz can be quite different. Depends on the wire size and the winding geometry.
Last edited:
There are more transformers poorly designed than well designed. Most people have no idea how to design a transformer that will have lowest possible AC resistance. The effect is audible.
Thanks everyone for explaining losses. I have narrowed down my search to Lundahl for flexibility and price, Onetics (truly excellent but wish it was potted for looks), and Monolith Magnetics (It's potted, great specs, and looks good).
I am having some difficulty understanding the wattage rating for Monolith though. Their B-8/5K push pull wattage rating is throwing me off. The PDF says “20 Watts rms in Class A1 mode ” and it weighs 5Kg. This seems awfully low as the Hashimoto 100W at 5K transformer is 5Kg, Lundahl LL1688 320W at 5.5K is 4.5Kg. Also, many of Lundahl's other transformers give around 100W with 5K impedance and weigh only 2.5Kg. These are all C-Core.
Products
Some other questions I have:
- Would adding UL taps compromise performance? I understand OPT transformers is the art of compromise.
- I might have Onetics or Monolith Magnetics custom wind an output transformer for 3.3K. Why do I keep reading that higher primary impedance's are better for performance? It seems that most 100W transformers for 3.3K have better frequency response and specs than their higher impedance brothers.
- How much is frequency response an indicator of quality? I know the Freed Citation II output transformers are famed for their frequency response. Although this might not always be the best indicator of quality...
I am having some difficulty understanding the wattage rating for Monolith though. Their B-8/5K push pull wattage rating is throwing me off. The PDF says “20 Watts rms in Class A1 mode ” and it weighs 5Kg. This seems awfully low as the Hashimoto 100W at 5K transformer is 5Kg, Lundahl LL1688 320W at 5.5K is 4.5Kg. Also, many of Lundahl's other transformers give around 100W with 5K impedance and weigh only 2.5Kg. These are all C-Core.
Products
Some other questions I have:
- Would adding UL taps compromise performance? I understand OPT transformers is the art of compromise.
- I might have Onetics or Monolith Magnetics custom wind an output transformer for 3.3K. Why do I keep reading that higher primary impedance's are better for performance? It seems that most 100W transformers for 3.3K have better frequency response and specs than their higher impedance brothers.
- How much is frequency response an indicator of quality? I know the Freed Citation II output transformers are famed for their frequency response. Although this might not always be the best indicator of quality...
Last edited:
Monolith Magnetics is definitively a step above for me.
20W Class A refers to the 300B Class A amplifier NOT to the transformer.
The transformer power rating is given below. It says saturation occurs for 30W @16 Hz. To compare with Lundahl you need see what happens at 30Hz (or see what the Lundahl does at 16Hz). So the Monolith M. will saturate at 105W @30Hz. So it is fair to say this is 90W @30Hz transformer (leaving minimal headroom like Lundahl). The Lundahl is 60W for 0.2 dB power loss. Much better frequency response for MM. Weight is also influenced by potting and enclosure.
It doesn't change the geometry. It's just a tap. However it might be more convenient to be a little bit flexible on the %.
It refers to the amplifier I think. Anyway a 3.3K transformer is not better than a 5K transformer by default.
O'Netics has other design principles. Especially at higher frequency they "characterize the sound of their transformers" by means of capacitive coupling and this usually doesn't not result in extended HF response.
For transformer is a good indicator. A transformer with extend HF is a good design. The contrary is not necessarily true. For an amplifier it's not , provided it can do 20KHz flat (or almost flat).
Quite possible Monolith uses core with very large cross-section area, thus reducing primary turns.
5K/20W - if I'm correct its for 300B PP, you can't get more then 20W from 2 x 300B.
Yet 5Kg weight for 20W only seem to be overkill for me.
1) UL transformers have very different winding layout compared to those wound for pure triode/penthode mode. In general yes, transformer with UL are higher class units.
2) Primary impedance - select you output tubes first. Typically 1.5K - 2.2K for KT120 / KT150, 3.3K - 4.3K - KT88 / 6550, 5K - 6.6K - 6L6/6P3S-E, 8K - EL84/6P14P.
3) HF limit alone don't tell all things, you need to know both leakage inductance and stray capacitance. Some PP amps, especially with deep global negative feedback (>=20 dB), don't do well if transformer leakage inductance is over 12mH. Anyway, if HF limit is over 130KHz, one can assume it could be a good unit.
Feedback is in fact extremely effective at reducing output impedance either in the form of the inherent output impedance of the output devices or indeed a real physical resistance in series with the output, it does not care. I regularly build valve amplifiers that achieve a 0.2 ohm output impedance in the midband using regular global feedback, rising slightly in the LF and HF extremes due to deliberate rolloffs in the open loop gain for stability reasons
You are mixing 2 different concepts for reasons unknown.
1) feedback **does** "things" , that's its job, why it's used at all.
Among other things, it flattens response, lowers distortion and lowers impedance (or rises it if set as current feedback).
Doing nothing my 4ss
2) now about resistive and hysteresis "lossy" loss, if I'm allowed to say so , whatever is turned into heat is gone forever.
A regular power transformer can easily be 95% efficient, since it can be optimizer without being tied by other considerations.
Being required to work at a fixed frequency (50/60 Hz) also helps simplify design.
Now an OT, specially a Tube OT, even worse if such tube is operating Class A so it will be continually wasting high idle current through it, will be WAY less efficient.
I smile at the quoted < 1dB losses quoted in some literature.
Well, if they *just* consider the AC/Audio component that might apply; if you consider the tube DC component, efficiency is abysmal.
Yes it applies when you consider the AC/Audio component.
Off course efficiency is small when you consider the "wattage" of a core for a power supply transformer and the needed "wattage" of the core for an output transformer.
Output transformers are supposed to work down to at least 20 Hz instead of 50/60 Hz.
Then, in case of single ended, efficiency gets worse because of the air gap.
A quality transformers needs ample winding space to apply interleaving and to keep DC resistances low.
These three elements mean that considerable "wattage" is needed.
My stock single ended output transformer for 300B has a core "body" of some 500 watts in terms of power supply.
I don't think I am mixing anything. Feedback cannot give you back power lost in the transformer. The reason why people confuse things (you too, it seems to me) is that they measure FR with small signals. Do it at full power and the loss will be there.
Last edited:
In my opinion loss in the transformer is a complex affair, and the source impedance has influence how the transformer performs.
For example, in a single ended amplifier, without global NFB, source impedance for the output transformer will be higher than with global NFB.
Using the same output transformer, the overall bandwidth for the feedback design will be better.
In other words, in the non-feedback design the output stage is not able to deliver the same amount of "power" because of the frequency response being worse.
Maybe a choice of "words", but nevertheless....
For example, in a single ended amplifier, without global NFB, source impedance for the output transformer will be higher than with global NFB.
Using the same output transformer, the overall bandwidth for the feedback design will be better.
In other words, in the non-feedback design the output stage is not able to deliver the same amount of "power" because of the frequency response being worse.
Maybe a choice of "words", but nevertheless....
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.