Note: Descriptions are shown in the official language in which they were submitted.
935.045
Z3~z
1.
¦ BACKGROUND OF THE I~ENTION
¦ Field of the Invention
This invention relates to digital-to-analog conver-
I ters. More particularly, this invention relates to monolithif
S ,I converters that are especially adapted for operation with
microprocessors, such a~ may be used in analog control systems.
D ~
li A wide variety of digital-to-`analog converters
~ I~ have heen availa~l now for some time. Such converters
ll frequently employ current sources which are selectively-
activated in accordancç with a digital input signal. U~ S.
I Reissue Patent No~ RE. 28j633 (Pastoriza) shows one highly '
I !l successful converter design ~f that t~pe. ~ more recent
l¦ design is shown in U. S. Patent ~o. 3,9~1~326 (Craven).
lS ¦~ Digital-to-analog ~onverters also are e~pl~yed in successive-.
~! approximation analog~to-digital cor~verters such as disclosed
~, in Canadian Patent No. 1,159,95~6 (Brokaw ek al~; the
latter converter particularly is advantageous in that it
incorporates inverted mode transistor circui.try (sometimes
. referred to as I L, for "integrated injection logic"),
, togeth~r with bipolar transistor circuitry, on the sam~
. monolithic chip.
-2-
935.045
~231Z
There have been pro~sa1~ for converters esp-cia11J
¦suited for use with microprocessors, as for example described
( lin the paper by SchoefE entitled "A Microprocessor Compatible
,Hiyh-Speed 8-Bit DAC", in the February, lg78, ISSCC Digest of
~Technical Papers, at p~ges 13Z-133.
¦ Although many proposals have been put forward,
none has provided a satisfactory converter capable of xequire~
pexform~nce, and yet suitably simple in design to be manufac-
),tured economically. Particularly, prior art designs have
llnot provided a monolithic converter including a reference
source and an amplifier to produce voltage output, all
'operable by a single supply, e~g. +5 volt. It is a principal
object of this invention to provide solutions to the problems
liencountered in achieving that goal.
15 ll SUMMA~Y OF T~E I~VE~TION
,I T~ nr~ wi t-l~ nnf~ ;mr~nr~nt~ ;~cnf~ nf the
invention, there is disclosed herein a digital-to-analog
converter in~orporating a superior ~uffer amplifier capa~le
, o producing an outpuk voltage between zero and some nominal
¦I value while operating from a single supplyO In accordance
with another aspect of the invention, there is provided a
simplified one-transistor current-source cell capable of
i operation directly with I2L switchi.ng logic, to permit a
2~ 935 045
.:ignificant reduction in part coun f the chip. I st~ll
another aspect of this invention, a converter is provided
( havlng a novel simplified reference supply for regulating
the gain of the current sources to stabilize the output-cur- I
rent of the converter. In yet another aspect o~ the invention,
a converter bias current network is provided having high
efficiency, znd thus reducing the requixed power while still
achi.eving excellent performanceO
.' 1,
~ Other objects, aspects and advantages of the
¦invention will be pointed out in, or apparent from, the
fvllowing detailed description of a preferred embodiment
of the inveniion, considered together with the accompanying
. ! drawings .
I I
i BRIEF DESCRIPTIO~ OF THE DRAWINGS
j FIGURE 1 is a block diagram illustrating the
llpxeferred embodiment;
l
ll
FIGURE 2A and 2B together present a detailed
~circuit diagram of the preferred embodiment,
i FIGURE 3 is a diagrammatic pxesentation of the
!'reference supply circuitry or the current sources;
1.
.
E~IG~RE 4 shows aspects of the buffer amplifier;
~} ~
935.045
~g231Z ~ '-
FIGURES 5 and 6 illustrate the functioning of the
.- bias current network; and
FIGURES 7A and 78 present a perspective section view¦
of the structure of a drive transistor used in the buffer
amplifier, and a schematic diagram of the elements of that
¦ transistor. I
.
DESCRIPTIO~ OF A PREFERRED EMBODIMæNT
Referring now to Figure 1, the microprocessor logic
control signals CS and CE are directed to a gate 20 which
¦con-trols the I2L data latches generally indicated at 22. The
11 latches are Lransparent to data on the data bus 24 when both
¦ICE and CS are àt logic ~0~'.. The input data is stored in the
latches when either CE or CS goes to logic "l".
. As shown in Figure 2, the data latches 22
~I comprise eight I L flip-^flops G20/G30 - G27/G37 (although
j for simplicity only the first and eighth are shown).
: ~ The out.put of the control gate 20, which swings around
l 1.2 volts, is applied to the bases of eig~t transistors
: I~Ql0-Q17 which act together with respective data input
: ¦I transistors Q~-Q7 to control the initially disabled I~L
1I yate pairs G~/GlO - G7/Gl7. Dependîng on the status o the
! input data, one or the other of the two gate lines 26, 28,
letc~, is pulled down to control the associated eight set/reset
~' l i
--5--
23~
,~35.045
lip-f].ops G20/G30 - G27/G37. If the ].eft-hand fli~ fl~p
~ section (G20-G27) is turned on, current will flow through the
( associated collector 30, etc., connected to the emitter of
a corresponding PNP current source transistor Q20-Q27. This
current is drawn from the ~ransistor emitter circuit, inter-
rupting flow ihrough the transistor to an R/2~ ladder network !
generally indicated at 32, and which functions in the usual
~way to provide an analog output current corresponding to the
~binary input on the data bus 24.
~eferring now in more detail to the current source
transistors Q20-Q27, the voltage between the common base line
.. . 34 and the emitter-resistor rail 36 is controlled by a regulat-
I ing reference source illustrated in simplified format in
Figure 3. This arrangement includes a band-gap cell of the
typa disclosed in U. S. Patent No. 3,8B~,863 (Brokaw), having
two transistors Q51 and Q52 with associated resistors R31 and I
R32 and operated at different current densities. As explained¦
in that patent, the collector currents of the two transistors ¦
are sensed by an error amplifier 38. In the present arrange-
ment, however, the amplifier output is directed to the Pmitter
resistor rail 36 to which is connectéd a reference resistor R8
and a reference transistor Q50. These elements, together with
another resistor R30, provide fee~back to the bases of tran-
llsistors Q51 and Q52, and drive the rail 36 to a volta~e
j,producing through R8 a feedbacX current which develops a
voltage across ~30 equal to the band-gap voltage (VGo~ e.g.
-v -
~ 9~5.045
essentially 1.205V. for silicon. This sets the current
through Q50 and all 8 DAC P~IPs at a temperature independent
lOO~A.
In the particular embodiment disclosed in Figure I -
2, the differential error amplifier comprises QS3 and Q54,
¦ and a common emitter output amplifier Q58. The pow~r supply
rejection of the loop is enhanced by common-mode feedback
I¦ through QS6, which fixes the collector base voltage of Q~l
¦l and Q52, and balances the error amplifier. A shunt regulator~
I generally indicated at 39 and of known design, biases the
l¦common base rail 34 at 1.2V.
. 1l .
; Referring now to Figure 4 together with Figure 2,
the output o~ tne R/2R ladder 30 is directed along a line 40
1~ to the output buffer amplifi~r generally indicated at 42
¦ and inclu~ing special features to accommodate the unipolar,
ground referenced output swing of the R/2~ ladder. This
amplifier comprises a ver~tical P~P di~ferential input pair
Q43/Q44 with Q44 serving to provide a feedback signal corre-
llsponding to the amplifier output voltage. The ladder signal
: 20 11 is coupled by Q43 up through a unique ~P~ current mirror cir--l
'¦cuit with iransistors Q34-Q39. The signal passes from Q36 and
~34 to Q37 which drives Q35/Q39 controlling Q40, a special
~,ldriver for ~he output emitter follower Q41 producing an out-
put voltage at the VoUT(E'~RCE) pin. This pin ~s connected
--7--
~ 312 935.045 '1
t~ the VOuT ~SE~SE) pin and ~; the Range Select pi~ for a
2.56 volt range. (Alternatively, the Range Select pin can
be connected to analog common, to provide a 10 volt range.)
The output voltage is fed back to the base of Q44, which
functions to halance the signal from the R/2R ladder.
. I .
One of the principal problems in achieving a
ground referenced output voltage swing in a single supply
converter results from saturation in the output stage when
the DAC signal level becomes very small, i.e. approaching zero,
thus limiting the ability of the output stage to drive the
output voltage down close to zero. In a nGrmal amplifier,
the conse~uent unbalance in the feedback causes the amplifier
to tend to overdrive, thereby further intensifying the satura
tion problem. This problem has been solved by a special cir-
cuit arrangement now to be described. I
Referring al50 to Fiyure 7, the output driver tran-¦
sistor Q40 includes the usual eptiaxial layer, a collector C
(n~), and a base ~p) having an emitter F (n+). In addition,
the base is formed with a second n~ electrode laheled C3, and
shown as a second emitter in the circuit diagrams of Figures I
2 and 4 This element is connected externally to the base of'
Q35. Normally, this element is not negative with respect to
the base of Q40, so that no current will flow. However, as
~23~2
~-- .
935 045
the amplif -r output voltage gets close to zero, th~s ~lement
begins to serve as an inverted-mode collector (thus being
labeled C3), and collects current so as to limit the base
drive at Q35. This prevents the saturation of Q40, and also ¦
controls the equilibrium points of the other transistors
associated with Q35. In effect, the additional collector
senses the onset of saturation in Q40, and operates through
an interior feed~ack loop including Q35/Q39 to prevent further
significant saturation while still developing the correct ,
output signal. The result is that overload of the entire
amplifiex is prevented, and the output voltage can be driven
down close to zero, even though only a single supply voltage
~is employed.
Tn ~rnr~nC~ with a further as~ect of the inven-
tion, the ou~put signal from Q40 prefera~ly is developed from
a separate collector C2 formed in a deep n* plug extending
down to the buried layer. This Kelvin connection avoids ~e
effects of the voltage drop across the internal resistance
hetween the conventional collector Cl and the buried layer.
.
jl R~ferring now to Figures 5 and 6 as well as to the
upper lef, hand corner o~ Figure 2~, the converter of this
invention includes a unique highly efficient biasing network
utilizing lateral P~Ps developing the relatively high level
currents reguired by the I L circuitry, the ECL control gate
2~ 1 20 and the reference supply for the current sources Q20-Q27.
_9~ i
23~ 93'; ~ 04~
Th~ sinale bias su~lv transistor Q75 is formed with split
collectors proportioned for the respective required bias
currents, and is deliberately operated in high level injec-
tion. Under these conditions, ~ and Ic (total) form a constant
product, i.e. ~ ICT = K. One collector 60 is connected to
a current m~rror 62 also connected to the base of Q75. In
more detail, as shown in Figure 6, collector 60 is connected . -
to the base of a feedback transistor Q73 the collector of
which is connected to the base of Q75. The collector 60 and
the emitter of Q75 are connected to a pair of transistors Q76
and Q77 having emitters with an area ratio of 1.25:1, to
establish that ratio cf currents therethrough~ The combined
current is delivered to the I L main injector rall. ~This
ail is the more positive of the two supply connections to
the I2L clrcuitry, and lS connected to all of the I L elements
indicated schematically as half~arrow emitters; the more
negative terminal is the buried layer.
With the current mirror feedback shown, the base
¦current IB for Q75 will be oC ICT M, w~ere CX~ is the
proportion of total collector current in collector 60, and M
~ is khe current mirror ratio. Since ~ is defined as ICT/IB~
I ;t can be developed that c~M = ICT/K, where K is the constant
previously described, resulting from the initial aevice design
characteristics Thus~ by properly ad~sting C~: and M with
~5 ~reference to the requirements of -the different circuit sections
-10--
.4 f
~35.0~5
fo~ bias urrents, i.e. by fixing the relative are~ :
emitters of Q76 and Q77 and the ~split collectors of Q75,
st~ble bias currents can be ob-tained for the individual
circuit sections. Moreover, as will be apparent from
S Figure S, the circuit arrangement makes use of all o the
current for biasing purposes, without loss. The circuit
in effect recovers both the base current and the collector
current of the feedback collector to be used as a current
source ~or biasing purposes. Figure 6 has been includea
alongside of Figure 5 to permit rèady comparison of the
block diagram presentation with the actual circuitry.
`
Although a preferred embodiment of this invention
has been described hereinabove in detail, i.t is desired to
lemphasize ~hat this has been for the purpose of illustrating
¦the invention, and should not be considered as necessarily
limitative of the invention, it being understood that many
modifications can be made by those skilled in the art while
: sti.ll practicing the invention claimed herein~
. I
CLAIMS: !