[Paul D. Spiegel]

When checking out transformers for my next project, I noticed that
Hammond has a new "A" series of push-pull output transformers. They are
more expensive than the standard series and I am curious if any RAT's
have experience with these and can report if they measure or sound any
better.

The minimal specs Hammond publishes look the same for the standard and
"A" series. The difference seems to be that the "A" series has a single
tapped secondary for 4/8/16ohms like the Dynace ST70 OT's and many
others. The standard Hammond PP OT's have two secondaries (4ohm and
4/8ohm) that are wired in various configurations to get the desired
secondary impedance. If I remember correctly, the 4ohm windings are
wired in parallel for 4/8ohm output and in series for 16ohms.

If the "A" performs better, then I may pay the extra amount. If this is
only about convenience, then I'll save $50 or so and configure the dual
secondaries as before.

Comments?

[Patrick Turner]

On Oct 12, 5:38*pm, "Paul D. Spiegel" wrote:
When checking out transformers for my next project, I noticed that
Hammond has a new "A" series of push-pull output transformers. *They are
more expensive than the standard series and I am curious if any RAT's
have experience with these and can report if they measure or sound any
better.

The minimal specs Hammond publishes look the same for the standard and
"A" series. *The difference seems to be that the "A" series has a single
tapped secondary for 4/8/16ohms like the Dynace ST70 OT's and many
others. *The standard Hammond PP OT's have two secondaries (4ohm and
4/8ohm) that are wired in various configurations to get the desired
secondary impedance. *If I remember correctly, the 4ohm windings are
wired in parallel for 4/8ohm output and in series for 16ohms.

If the "A" performs better, then I may pay the extra amount. *If this is
only about convenience, then I'll save $50 or so and configure the dual
secondaries as before.

Comments?

There has been a trend amoung tube amp manufacturers to avoid the use
of multiple secondary windings which can be arranged in various ways
to obtain specific "load matches". Such arrangements can only be
altered by someone technically qualified and who can use a soldering
irn properly.

So what happens so often is that dumbass owners try to re-arrange sec
windings and invariably make a **** up and then they get smoke and
blame Hammond or whoever made the amps.

Amp makers seem to have universally avoided any provision of a load
matching switch which makes the alterations required without a
soldering iron but in away which involves the use of a really
blameless type of switch which really cannot ever be found amoung
switchmaker brands at any sort of reasonable price. There is almost no
demand. A really good switch might be the same price as the OPT
itself. Some ppl have asked why not have a 12 pin plug and socket for
3 different possible inserted positions to get 4, 8 and 16 ohm
matches. But ppl lose the plugs, and then you are ****ed again and
makers loathe the extra expense on what is always cheaper.

The tube amp industry has responded to the idea that having variable
or selectable load matching is a good feature and good for sales.
Amp makers seem to think owners like to change load matching to get
the best sound and the most possible power.
And amp makers know buyers of their amps are entirely ignorant of
basic electronic issues of any kind. So the amp makers try to please
the buyer by giving him the most features which are least bamboozling
and this means 4 terminals at the amp rear, Com, 4, 8 and 16.
Sometimes there is only Com, 4 and 8.

Hammond has hopped on the bandwagon to give us a TAPPED SECONDARY and
if we want we may have terminals at the amp rear labelled Com, 4, 8
and 16 ohms.

Ppl need to consider the difference in properties between multiple
arrangable sec windings and a single winding with taps.

With multple windings there should be equal current density in all sec
winding wires and no un-used portion of any secondary windings. This
means the % winding resistance losses are constant for whatever load
is connected when the windings are arranged for the load meant for the
arrangement. It also means the leakage inductance remains constant so
HF bandwidth remains constant.

With a tapped sec, the minimum winding losses occur only when all the
sec is connected to the load meant for it.
Suppose that is 16 ohms. There may be a tap of 0.7 of the total sec
turns for 16 ohms along the winding and this is for an 8 ohm load.
There may be another at 0.5 the total turns and this tap will be for 4
ohms. When 4 ohms is used, only 0.5 of the total sec turns are used
and the remaining 0.5 have no current at all. So the sec winding
resistance losses are much higher with a 4 ohm load connected. Since
0.5 of all sec turns are used the leakage inductance may be much
increased and the bandwidth reduced so that stability with NFB could
become a problem.

In general, multiple sec windings allow us the possibility of making
the best possible OPT and the tapped sec will always sacrifice
optimization for convenience.

I now happen to be working on a pair of Dynaco Mk4 monobloc amps which
has OPTs with tapped secs. I have had experience working on Mk6 mono
blocs and these old amps from the 1960s had tapped secondaries.

They seem to have a winding arrangement of P and S section
interleaving where there are TWO sec sections interleaved with
probably 3 primary sections.

The two sec sections appear to have a maximum of N turns each meant
for 16 ohms and the the two sections are paralleled and bandwidth is
acceptable because of this minimum of interleaving. Better OPT would
have 5 S sections and 4 P sections but you'd have to inspect my
website to understand this better.
Anyway, each S section of the Dynaco OPT has taps for 8 and 4 ohms and
so although there appears to be ONE sec winding with taps there are in
fact more than ONE sec winding and these equally tapped windings are
paralleled so that the interleaving pattern is not reduced adversely,
and nor is the HF response much reduced. Winding losses do increase,
but then even it winding losses are as high as say 12% the sound may
still be very good because winding reistance is just only resistance,
and resistance itself is not a non linear distortion causing circuit
element.

Some makers might put two sec sections into a given OPT each of 0.5N
turns, and connected in series to give a match to 16 ohms. There may
be a tap at 0.7N turns for 8 ohms while connection to just one lot of
0.5 turns is a match for 4 ohms.
When 4 ohms is used there is an unused section of sec winding and the
interleaving changes from say 3P : 2S with 16 ohms to 2P : 1S and this
makes HF response much less.

So there is an optimal way of designing tapped windings and by careful
examination of Dynaco OPTs I can say that the designers went to some
trouble to convince the accountants that the method employed was
necessary.

I have almost no idea how exactly the windings are laid up in the
bobbins of Hammond OPTs.
Hammond do not give anyone a bobbin winding diagram showing exact
sizes of wire, insulation and turns so that those of us who think all
the fine details really matter could check it all out properly. Is
Hammond management really ashamed of what we might find out? Do they
like to keep trade secrets buried and away from those who might copy
their work? Who knows?

Hammond are quite cheap. Hooray for that. Try making an OPT yourself,
and suddenly you'll see that Hammond are a real good deal. if you want
better, better is available, and maybe you have to go to Plitron or
Lundahll or Sowter or some other fancy brand and you'll wait months
and pay 3 times the Hammond price.

I have used Hammond 1650P for replacements of the ****ing awful
quality of Chinese made OPTs in Jolida 502 amps and where the Chinese
crap has developed shorted turns in the primary, thus requiring OPT
replacement. The Hammond drop straight into the Jolidas and the bolt
holes match. Load is 6k6 primary load which is better than the
original 5k0 load used by the Chinese who are so recalcitrant about
learning how to make a decent tube amp. But the leakage inductance of
the Hammond appears to be no lower than the original Chinese crap, and
IMHO, way too high at over 50mH referred to the primary which makes
getting unconditional stability with NFB somewhat difficult to
acheive. I don't know if the type A Hammond are any better but let's
assume the leakage is way too high just like the older range of their
OPT.
However, I managed to get the Jolidas to become good sounding and
reliable with Hammond OPT despite the shortcomings of the OPT.

So to a large extent it ain't what ya got that matters as much as how
ya use it!!

With the old style Hammond 1650P you can have Com, 4 and 8 ohm
terminals at the amp rear which means you have paralleled two 4 ohm
windings for 4 ohms and the 8 ohm is gained because one of these
windings has extra turns to get a match for 8 ohms. In today's world
there is never any need to match to 16 ohms.

Now, the "anode to anode primary load" of 6k6 is gained when you have
4 ohms connected between Com and the 4 terminal.
Similarly, the RLa-a = 6k6 when there is 8 ohms between the Com and 8
terminal.

But there is a third value of speaker load load which can be used to
get a primary load of 6k6 if we connect between the 4 and 8 terminals.
Between the two, the turns used are 0.285 of the total needed for 8
ohms so the load match between 4 and 8 becomes 0.65 ohms.
One might say this is useless, but suppose that is where someone
connects a 4 ohm speaker. The load reflected to the anodes becomes 40k
and max PO becomes perhaps 7 watts only but it is 100% class A1 and
THD is extremely low.
Not a bad thing if you have horn speakers - and if your amp stays
stable.....

However, most ppl would not explore possibilities using the 4 - 8
connection.

But they may well become confused. With only TWO output terminals at
the rear of any amp there is minimum confusion possible. If the user
manual ( which rarely ever gets read ) says the speaker should be 8
ohms, then use only 8 ohms or more ohms but of course ppl will try to
use 4 ohm speakers, or speakers with known reputations for killing
amplifiers such as the AR9.

If there are 3 terminals, Com, 4 and 8, it invites smoke. The owner of
4 ohm speakers might think 4 ohms is less ohms than 8 ohms and
therefore "an easier" load to drive so they plug the 4 ohm speakers
between Com and the 8 ohm labelled outlet.
The impedance ratio of the OPT is 6k6 : 8 ohms or 825:1, so 4 ohms is
transformed to a load of 4 x 825 = 3k3 and this may cause the OP tubes
to overheat and cause collateral damage in the amp.
But let us say the owner has 8 ohm speakers and he makes the correct
assumption that 8 ohms is "easier to drive" than 4 ohms.
He plugs the 8 ohm speakers between Com and 4 where the impedance
ratio = 6,600 / 4 = 1.650:1 and the anode load the tubes experience
becomes 1,650 x 8 = 13k2, or twice the load ohms of 6k6. With the
higher load the tube operation will move from the normal class AB with
6k6 to nearly all class A. Power max at clipping will fall from say
35W to say 25W but it will still be enough. Damping factor will double
and THD will be reduced and bandwidth increased, so its all positive
compared to using a 4 ohm load plugged into the 8 ohm outlet.

In conclusion I would therefore say that you would never need to use
the type A trannies and I don't see why extra should be paid for it.

In amps have made for customers I will often have only 2 output
terminals and for a nominal match for 5 ohms. For my SE amps this load
is the one where the amps maximum PO is possible. All power is class
A1. So load tolerance is in practice quite good if low average power
levels of less than 10% of maximum possible is used then any load
between 2 and 20 ohms can be used.
In PP amps I like to give a single match to 5 ohms, but it is with 3
ohms that the maximum PO in class AB1 is obtained.
Therefore any load value between 2 and 20 ohms will be tolerated at
the low power level used by audiophiles.

In most SE or PP amps sold these days the maximum PO is obtainable
when the load used equals the labelled outlet.
So if the outlet says 8, the two ohm load is a long way different and
the amp struggles, maybe it burns.
Many mass made PP amps have outlandishly low anode loadings of say 3k0
and the tubes are operated with fixed bias and whipped into making as
many watts as possible at clipping and the results are quoted as
"marvellous headroom" but the horse is almost dead from the whipping
it gets. The tubes teeter on the edge of thermal collapse and runaway
and DF is poor and THD is high. Many hi-fi amps made have very poor
load matching.
To exacerbate a bad trend of amp makers addicted to producing crap is
the tendency of speaker makers to produce ever lower sensitivity and
impedances for their speakers. So we have many speakers which may be
nominally "6 ohms" but which have a large portion of the main power
range between 100Hz and 1kHz where impedance dips to 4 ohms or lower.
This is especially so with 3 way or 4 way speakers and there are
second order crossover filters whose input impedance becomes lower
than the speaker nominal impedance at the crossover frequency.

So now you can appreciate why I like to give owners a match for 5
ohms, and tell them "use anything you like."

I have anticipated that owners will indeed use any damn thing they do
like, even ancient AR9 to just over 1 ohm at 120Hz and which are
fairly insensitive.
Then there are the range of electrostatic speakers which may be a
problem.
But no matter what is connected to any of my amps the stabilty will be
fine and as long as average level is under 10% of clipping level no
harm will be done.

If 16 ohms speakers are used, chances are they may be more sensitive
than 8 ohm types so driving these from an 8 ohm outlet of even 4 ohm
outlet usually gives excellent results and a dedicated 16 ohm outlet
is rather un-necessary.

Patrick Turner.